Skip to content
Notifications
Clear all

[Sticky] USB-C Microphone (official topic)

819 Posts
154 Users
172 Reactions
65.8 K Views
(@ldoppea)
Posts: 30
Eminent Member
 

Thanks for the link to Audacity documentation. I found the silent monitoring mode but it seems not very precise, also I didn't found the "Enable audible input monitoring" option, I'll try again with it enabled.

 

 and I'll try to check in often so you're not waiting

Thank you but don't worry, I'm currently in a 3 day conference, so I won't be able to test this until Sunday 😢 

Next thing to try is to check the resistance between pin 1 and pin 8 on the THAT1512.

Just to be sure: should I check the resistance on the chip in the circuit? Or should I unplug it first?

 

This would be the next thing to do in fact. I would suggest you remove the gain setting resistor/switch completely and attach something that will drive a small headphone direct to either of the inputs. It's important to "inject" the signal via a capacitor though as both inputs have a DC potential on them

I'm not sure to understand that part, especially the following part: "attach something that will drive a small headphone direct to either of the inputs. It's important to "inject" the signal via a capacitor though as both inputs have a DC potential on them"

Do you mean I should plug the +/- pins from a Headphone's jack connector to the "gain" pins instead of the rotary switch? I'm not sure this is what I should understand as headphone is an output devices and your are talking about inputs?

Also I just don't understand what "injecting a signal via a capacitor" means in that context.

 
Posted : 01/02/2024 12:09 am
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

Pin 1 and 8 on the 1512 (at least at the socket, so yes, better to remove it) and remember to always let the circuit "cool down" so the caps discharge before sticking a resistance meter in there. Sorry, I could have made that clearer.

1 and 8 are the gain pins - if there is an infinite resistance there (i.e. the gain switch is uplugged) the device works at unity gain. That means the voltage at the output is the same as the voltage at the input.

When I'm talking about the headphone output, I'm referring to another piece of audio - ideally one you don't care about. I could be an old CD/DVD player, boom box, radio you name it. The headphone outlets on theses (especially the older ones) are AC coupled so often quite robust and provide enough voltage to drive the ADC. 

This is a good way to see what might be ailing your circuit.

Now the "correct" way to do this is with a signal generator, but this route saves getting one. You can get them on eBay quite cheaply using some old function generator chips and their clones. But it's all more expense. 

When I mention putting a capacitor in the test harness, it's because the design puts + or -15 volts up the signal lines (a form of phantom power) and some devices won't have a capacitor at their output which means there is a potential to blow that up.

The capacitor (which needs to be rated at at least 16v, but 25v is better for headroom) and ideally isn't an electrolytic - protects the working device from a sudden voltage at its output stage. Many devices are AC coupled, but not all are and you know how Sod's law goes, right?

The idea behind this is to get a nice beefy signal - ideally around 1V p-to-p into the 1512 and see what comes out the other side. I'm guessing this won't work either but if it does, it might point to noise causing the device to saturate.

You can also check the voltage on the 1512's power pins in this case - if they have jumped to 15 volts or more, with 0dB of gain and drop to I think you said 13 volts as you turn the gain up (but without the capsule attached) that suggests saturation from noise. Never a fun thing to deal with esp. as that noise is well outside of the range we can hear and I fully expect outside the range the ADC can cope with.


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 02/02/2024 12:58 pm
(@ldoppea)
Posts: 30
Eminent Member
 

I just made the measurement. I confirm I get infinite resistance between pins 1 and 8. Just to be sure, I identify pins based on this pinout scheme:

image

 I'm referring to another piece of audio - ideally one you don't care about

I'll try to find one, I may have an old ipod somewhere, hope its battery still work.

 

Now the "correct" way to do this is with a signal generator, but this route saves getting one

Would an Arduino board be useful for that?

 
Posted : 04/02/2024 4:48 pm
(@robert)
Posts: 6
Active Member
 

@dnamein Hi man, sorry to get back this late to you, I havent checked the forum in a while as I never got a reply to my question.

Glad it finally works for you, however, for you the same 'problem' persists that you need an external power supply, right? Kind of strange that we both face this issue...

 
Posted : 04/02/2024 8:01 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

@robert I'm looking at dual supplies for the V2. Part of the problem is that there are some great ways to generate a ground loop via the disparate earths. The problem with a USB ground is we don't know if it's connected to mains ground or is isolated. It usually will be in a laptop but I expect it's probably connected in a desktop machine.

I blew it (derp) myself by sharing a ground from the isolated NMA0515. So I'm making two separate signal paths - a 5V one and a high-voltage (48V) one. I'm testing this thing to death now which is why it's taking so long and so much money - but I'm not going do drop designs that won't work.


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 05/02/2024 2:38 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

@ldoppea Right pin layout but be aware that's not the INA layout we have on 1510/1512 or any one of the other DIP8 instrumentation amplifiers. I'll try to refer you to pin numbers where possible since they are easy to measure.

OK, the impedance across the two input pins is measuring the amplifier's internals -but that's OK, we want the THAT to run at "unity" gain - which means that it's output essentially follows what the input sees.

Now this isn't a lot of use for a mic amplifier but it's a great way to verify the circuit works by injecting a signal into the input pins (one of them anyway).

I guess (based on your choice of chip image above) that you're familiar with Arduino (the I2C bus bins are a dead giveaway). 

The Arduino eco-system is huge now but if you could produce a tone of about 1KHz square wave at about 1V peak - you might need to use a potentiometer to get the swing down. The ESP32s if you have one of those can produce a nice cosine wave which is far less harsh on your ears.

A "headphone" source is the best way to do this. There are three tests we can do.

Connect the outer screen (braid) to the 0V which is pin 4 on the THAT but it's probably best soldered to one of the common "rails" since you don't want it flapping around in the wind and shorting something.

Ensure the signal is playing, set your equipment to capture (the card Matt suggested will flash while it's recording) and then "inject" the signal from your music player, etc. into Pin 2 OR Pin 3 and see what happens.

If you apply the signal to both pins at one, the output should go to zero. This is by design and nothing to worry about. The main thing is that pins 2 and 3 are putting a signal out. 

 

 


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 05/02/2024 6:54 pm
(@ldoppea)
Posts: 30
Eminent Member
 

Hi,

I finally got time to make some tests using my old arduino board.

I used the tone() method to generate a 1KHz square wave at 5V peak. I did not find how to directly generate a 1V wave, so I used a resistor based voltage divisor and I finally got a 1V output (0.9V to be exact) when setting Arduino's pin to full 5V, and ~0.5V when asking to output a square wave (which seems to be normal if I understand it correctly)

When trying to plug the square wave on pin 3, nothing happens on Audacity. Like for my previous tests, there is some noise at the exact moment I touch the pin 3, but after that Audacity shows no sound. I get same result for pin 4.

Connect the outer screen (braid) to the 0V which is pin 4 on the THAT but it's probably best soldered to one of the common "rails" since you don't want it flapping around in the wind and shorting something.

I think I misunderstood that line, I tried to plug the GND to pin 4 (I understood "braid" to be the braidboard), but this makes the arduino board freeze (I suppose it goes security mode). Hopefully this didn't break the board, I made some new checks and I still get the correct voltages on the pin that generates the square wave.

 

Here is where I plugged the Arduino board

image

 

 

 
Posted : 10/02/2024 5:48 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

Ouch. Not surprised the Arduino board froze - pin 4 is minus 15v.  That was my error - I was working on a single rail design at the time with a 0v on pin 4. Typical of 8 pin designs. But I apologise for giving you bad info. 

They are pretty tough as a rule but that's putting a large demand on the output driver which, I assume, brought the LDO to its knees and shut the processor off.

Which (I hope!) shut the outputs down and protected the MOSFETs from getting fried.

Mind you most devices would be in a lot of "pain" if you exceed the voltages by such a large amount (in my day that sort of mistake would be "flashy-sparky-ouchy-smelly") 😉

Although there's no official announcement for now, I just completed the first revision a complete 5V solution (total USB power, no expensive chips and not much soldering). I don't expect this one will come up to the standard Matt reached with this one but it's simpler for beginners. I'm not going to publish it until it comes back with the other orders which aren't going in for about a week or two. All complete new designs from the ground up.

BTW, very clear graphic work there. Makes it easier to see where things are going.

This post was modified 9 months ago 2 times by marcdraco


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 10/02/2024 7:57 pm
(@ldoppea)
Posts: 30
Eminent Member
 

But I apologise for giving you bad info

Don't worry, I'm infinitely grateful that you help me

Also I have my part of responsibility, I was not sure to understand and instead of asking you I searched on google the meaning of VSS (what is displayed for pin 4 on the schematic I shared last week) and got as result something like "VSS is the lower level, often the 0V GND", and I did not check what was behind the "often" part 🙃 

That being said, I have some questions about last experiment.

I forgot about that part, but was I supposed to do this test with the electronic board inside of a metallic box?

If no, my understanding from that test is that there can be 3 sources of errors. Either the THAT is not working, or my acquisition card is not working, or the problems is somewhere on the board between those two components (i.e bad welding). 

Last time i used the acquisition card (a few month ago) it worked correctly, but today I have no functional jack microphone to test it. Do you think it is worth to invest time finding a way to test it? Would it be possible to use the Arduino board here again to inject a signal directly in the jack connector?

 
Posted : 12/02/2024 5:58 pm
marcdraco reacted
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

That's gracious of you and of course you're welcome.

Generally, "Vss" is the most negative leg of a FET-based IC. The most positive is the drain do we get "Vdd". Bipolar devices use Vcc (collectors) and Vee (emitters).

I just caught a very subtle error in one of my new mic boards which would (had I not spotted it) have shorted the supply rails. It sounds impossible but I "forgot" to put the circuit voltages in order so as the voltages get lower, the tracks appear lower in the diagram.

The nature of the supply I had in this is strange: having two positive voltages and a ground. That's all. And when I was drafting it out I just forgot and connected it to the lowest point in the circuit. A case of familiarity breeding contempt! I need to get some more SPICE models for KiCAD so I can check my drawn circuit works they way I think it does!

It's unlikely the acquisition card has failed, but you're correct in thinking that you can use the same "audio" source for your tests. Consumer devices tend to be more robust in some regards because you never know what someone is going to plug in! Professional stuff has protection too - just a lot more of it (but for the same reason).

As for the experiment, at "unity" gain your circuit won't (or shouldn't!) but putting out any detectable signal, even with pins 2 and 3 "waving in the breeze" as it were.

A such a low gain, almost literally 1:1 in fact, the only signal present when there's no input will be the THAT1512's internal noise which isn't measurable in by the acquisition card as it's the noise in that device swamps the noise from the 1512.

That means you can work without any fancy screens, making life a whole lot easier for this stage of the game! [Hiding this in your reply] I've just dropped the Gerbers and BOM for a simpler 5V only design, announcement elsewhere on the site AND four brand new mic head designs with (in simulation) response well into medium-priced professional. [See most people will just skip over ... :)]

But yeah, that should give us an idea where the problem is. it's always best to break a problem into two (equal) halves and figure out which half is broken. Then you break the broken one into two problems and so on. It's called divide and conquer and will save you hours of tearing your hair out.

Marc


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 12/02/2024 8:03 pm
(@frankenbanana)
Posts: 1
New Member
 

I've been wanting to build this project for a while now and just decided to pull the trigger recently, I finished it however when I plug in the microphone although the computer detects it there is no sound detected whatsoever. for simplicity I am using a usb a connection and connected the power siphon to the usb audio interface pcb as opposed to the usb connector to make it easier although I dont believe this to be the issue I'm trying to be as transparent as possible. I also soldered together the connections that would be used as the volume control knob so as to remove any variables possible that werent absolutely necessary for showing that the pcb works in that it can actually create sound at all which it isnt right now. other than that I have tried my best to do this identically as instructed and still do not know where my error lies.

IMG 7560
IMG 7557
IMG 7555
IMG 7559
IMG 7554

 

This post was modified 9 months ago by Frankenbanana
 
Posted : 13/02/2024 9:58 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

Hi there and welcome to the forums.

First thing I've noticed, you have a screw through the stripboard which might be causing a short. It's worth double-checking you haven't accidentally shorted a couple of tracks with a copper sliver or similarly not fully cut the track.

Shorting the volume connections will push the THAT1512 beyond its published design limits, which is harmless but it also means that if it's oscillating (which is highly likely) it might be "saturated" at a frequency that we cannot hear and that the digitizer can't reproduce.

It would help if I know what sort of test gear you have.

I assume you have a multimeter but a scope and a function generator would help.

There are many things we can test to see what might be going on but a lot of it requires some test gear, if that makes sense.

For instance what voltage is at the various pins of the THAT1512 (with respect to ground/0v), especially the output: pin 6 at the device itself so we don't get a read from an output capacitor.


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 20/02/2024 10:14 am
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

Hi there and welcome to the forums.

First thing I've noticed, you have a screw through the stripboard which might be causing a short. It's worth double-checking you haven't accidentally shorted a couple of tracks with a copper sliver or similarly not fully cut the track.

Shorting the volume connections will push the THAT1512 beyond its published design limits, which is harmless but it also means that if it's oscillating (which is highly likely) it might be "saturated" at a frequency that we cannot hear and that the digitizer can't reproduce.

It would help if I know what sort of test gear you have.

I assume you have a multimeter but a scope and a function generator would help.

There are many things we can test to see what might be going on but a lot of it requires some test gear, if that makes sense.

For instance what voltage is at the various pins of the THAT1512 (with respect to ground/0v), especially the output: pin 6 at the device itself so we don't get a read from an output capacitor.


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 20/02/2024 10:15 am
(@ldoppea)
Posts: 30
Eminent Member
 

Hi,

Sorry long time without news.

I did not test the acquisition card yet, but I made some research to be sure to test it correctly (my main question is there is a max voltage that I should inject in the jack entry)

BUT by doing this research, I may have found what is the problem (but i'm not sure).

In order to use the acquisition card without breaking it, I tried to use an old male jack plug from one of my previous projects. And I realized that it may not be compatible with my acquisition card.

Based on the card documentation, I should use TRS jack (or a TRRS jack, but I don't use this entry)

image

And it turns out that my jack is a TS one 😓 

image

But I'm not sure if this is the problem. I found some articles that seems to say that it is possible to plug a TS male jack into a TRS entry. But the documentation says that we shouldn't do that. What do you think about this?

This post was modified 9 months ago by Ldoppea
 
Posted : 20/02/2024 9:00 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

I know the very card, I have one here!

Ok, so the three plugs are like this:

TRRS = Tip, Ring, Ring, Sleeve (stereo with microphone)

TRS  = Tip, Ring, Sleeve (stereo/balanced)

TR    = Tip and Ring (mono/unbalanced)

TR is the original form from (used in telephone exchanges in the 1 1/4" version and we inherited a smaller version which is reliable, cheap and robust but cannot be waterproofed (apparently) which is why it's been removed from higher end phones.

The sleeve is always your ground reference line (0v) from which all your other signals are referenced. You can ignore the mic input for now as that's effectively unused in a *stereo* (TRS) socket. You don't tend to see balanced lines on 3.5mm plugs as they use beefy cables that won't fit. They are also insufficiently robust for professional use.a

So what happens when you plug a mono input into a stereo socket?

Not a lot as it happens.

"They" say you shouldn't do it because its best practice to stick with a stereo TRS jack into a TRS socket. As far as inputs are concerned, one of the channels will be grounded. The easy way to to remember is the the Ring = Right and that channel will be grounded (silent). The tip connects your signal to the left channel. 

A general rule is you can always ground an input but never an output.

You'll be able to test this with your Arduino quite easily.

Max voltages for input signals are usually determined by the rating of the input capacitors (they'll typically be the lowest voltage they can get away with - so likely 6.3V on a 5V USB device).

For completeness, I should mention that most (by no means all!) modern consumer electronics amplifiers have output short circuit protection. Many operational amplifiers have this too - you'll see this mentioned as "output short circuit current" and the duration, often this is unlimited as the protection kicks in and limits the output so the chip doesn't burn out. A strange design inclusion perhaps but it adds very little to the cost of these things which are produced by the million - and something that's saved many a designer (including professionals) from the infamous "magic smoke".

For the THAT1512, a short circuited output to ground is instantly limited to 35mA and it'll stay like that until the fault is rectified.

For comparison, my go-to dual Op Amp, the NE5532 ranges from 10 to 60mA (38mA typical).

That's not that we should go around shorting these things but we don't necessarily know in advance if this thing is going to end up shorted by a circuit fault or an idiot. 😉 I recently saw an experienced repair guy on YouTube connect his PSU up with the "wrong" ground (causing an earth loop) if that had input protection, it failed: badly!

I think that probably answers your questions but keep 'em coming and tell me if I miss anything.


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 20/02/2024 9:54 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

I know the very card, I have one here!

Ok, so the three plugs are like this:

TRRS = Tip, Ring, Ring, Sleeve (stereo with microphone)

TRS  = Tip, Ring, Sleeve (stereo/balanced)

TR    = Tip and Ring (mono/unbalanced)

TR is the original form from (used in telephone exchanges in the 1 1/4" version and we inherited a smaller version which is reliable, cheap and robust but cannot be waterproofed (apparently) which is why it's been removed from higher end phones.

The sleeve is always your ground reference line (0v) from which all your other signals are referenced. You can ignore the mic input for now as that's effectively unused in a *stereo* (TRS) socket. You don't tend to see balanced lines on 3.5mm plugs as they use beefy cables that won't fit. They are also insufficiently robust for professional use.a

So what happens when you plug a mono input into a stereo socket?

Not a lot as it happens.

"They" say you shouldn't do it because its best practice to stick with a stereo TRS jack into a TRS socket. As far as inputs are concerned, one of the channels will be grounded. The easy way to to remember is the the Ring = Right and that channel will be grounded (silent). The tip connects your signal to the left channel. 

A general rule is you can always ground an input but never an output.

You'll be able to test this with your Arduino quite easily.

Max voltages for input signals are usually determined by the rating of the input capacitors (they'll typically be the lowest voltage they can get away with - so likely 6.3V on a 5V USB device).

For completeness, I should mention that most (by no means all!) modern consumer electronics amplifiers have output short circuit protection. Many operational amplifiers have this too - you'll see this mentioned as "output short circuit current" and the duration, often this is unlimited as the protection kicks in and limits the output so the chip doesn't burn out. A strange design inclusion perhaps but it adds very little to the cost of these things which are produced by the million - and something that's saved many a designer (including professionals) from the infamous "magic smoke".

For the THAT1512, a short circuited output to ground is instantly limited to 35mA and it'll stay like that until the fault is rectified.

For comparison, my go-to dual Op Amp, the NE5532 ranges from 10 to 60mA (38mA typical).

That's not that we should go around shorting these things but we don't necessarily know in advance if this thing is going to end up shorted by a circuit fault or an idiot. 😉 I recently saw an experienced repair guy on YouTube connect his PSU up with the "wrong" ground (causing an earth loop) if that had input protection, it failed: badly!

I think that probably answers your questions but keep 'em coming and tell me if I miss anything.


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 20/02/2024 9:55 pm
(@ldoppea)
Posts: 30
Eminent Member
 

Hi,

I was not sure to understand everything 2 days ago, but yesterday I made some new tests and then today, after reading it again I think I understand it more 😋 

So I made a few test to understand what works and what doesn't works

First I tried to inject a signal with arduino in the jack input. I just plugged the acquisition card to my computer, and with the arduino I tried to inject the same 1v sinusoid as on our previous test. Nothing really relevant. A lot of noise happens in Audacity at the moment I put the contact, but then nothing.

I was then frustrated because my knowledge did not allow me to understand if this was a proof of the acquisition card failure, or it was just a proof (i forgot to ask you if your previous secret message did catch the attention of some people?) that I was doing some derp things with no sense and still expecting some result  🙃 

The only way for me to be sure of this was to find a working earphones with mic. Took me 1h searching through all my old electronic boxes, but I finally found one that still work. And the result is that the acquisition card is working correctly.

Those earphones are TRRS (this is the only viable connector for stereo output and mic input I suppose?)

Here are the result:

  • Earphones plugged to my computer -> it works 
  • Earphones plugged to the acquisition card on the black input (TRRS) -> it works
  • Earphones plugged to the acquisition card on the pink input (TRS) -> it does NOT work
  • The TS connector I used on my initial circuit plugged on the pink input (TRS) and its wires in contact to the earphones' jack -> it works (but with a lot of noise probably due to the poor contact)
  • Same thing but the TS connector plugged to the DIY circuit and I tried to input the earphones' jack into the OUT pin of the DIY circuit (cf schema) -> it does NOT works (but it was really hard to keep the contact in place so i'm not sure about any conclusion)

 

image

 

Since yesterday I realized that I should have tried the same thing without the THAT plugged or without the DIY board alimentation in order to cover every combination. And I have to find a better way to do the contact on the earphones' jack (ideally without having to solder anything)

 

Also one important thing to note. I don't know why, but at some point during my tests, the acquisition card would stop working when pluging its alimentation to the DIY board's V+/GND connector. This was the first time this happened. Also when in this state, my computer's would alert me that the USB port was drawing too much power.

I unplugged everything, rebooted my computer (otherwise the alert would stay), tried to re-plug the acquisition card alone, and it worked well. I then tried to plug the V+/GND connect, then same result.

I got the same result after a third try, then made a pause, and when I came back and made a last tried, then I succeeded to make it work again without any noticeable change.

I still don't know what happened, maybe you will?

 

For now I don't have any conclusion about those test, except the fact that the acquisition card is working. I will try to do the test without the THAT nor the alim tomorrow.

Until then, do you have any suggestion? Are there other tests I can make?

This post was modified 9 months ago 2 times by Ldoppea
 
Posted : 22/02/2024 11:08 pm
marcdraco reacted
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

Oh, I wish other people would record their tests as well as you - makes it so much easier to figure out what's going on. Don't lose faith, you're doing well!

When you use the word "alimentation" do you mean "power"? I'll assume that's what you mean given that it seems to be in context. I'll use power in my replies for the native speakers who might get tripped over that.

Many headphones/earphones for GP use are designed for mobile phones since that is the main market, hence they are TRRS to allow for stereo sound and a separate microphone so that people can talk hand's free.

There is an accepted colour code devised, I think by Microsoft some years back. I forget most of it but green is headphones or "line" out and the pink is mic in. Black is probably rear, but that could be centre too and my cat is using my good hand as a... I'm not entirely sure but he won't give it back. Rather slows me down until he gives it back.

Posted by: @ldoppea

Those earphones are TRRS (this is the only viable connector for stereo output and mic input I suppose?)

In a nutshell, that's exactly it.

Now here's where it gets a little tricky. The mic input *also* carries a DC voltage (about 1.5V) to power the FET in the cheap but effective electret mic in those headphones you can pickup at the dollar store. It's called "bias" for simplicity but it's more like a combined power in and signal out, a poor-mans 48V phantom power if you will.

A standard mic input designed for plug in electrets will be TS only and that means the supply voltage (also called "bias") won't reach the correct ring on the TRS or TRRS to supply power to the microphone. This isn't a short circuit because there's a current limiting resistor in the bias chain, usually 2k2, that stops anything getting damaged.

You can "test" this socket using an electret microphone with an internal FET and it's possible to wire the JLI capsule to do this (the details are on JLI's datasheet but while it's straightforward, the quality suffers a lot.

To be technical, the FET is operated as a common source gain stage of about 20dB (but nowhere near as predictable) as a source follower. Don't worry about looking that up, it's pretty heavy on math and semiconductor theory. (Matt's design sidesteps some of that that by using a phase splitter, but the quality way to do it is a source follower in a negative feedback loop; with a considerable increase in complexity. That's the aim of my final "head" amp - but it's stuck over in China because I made a mistake on the size of a couple of small parts.) Naturally this all costs more too - but it's academic right now.

Put simply, don't use the microphone (pink) port on that card as it won't work. Matt's board is specifically designed to drive the green (LINE) input on these devices.

The difference is that the pink port supplies a bias voltage and a signal of a few 10s of mV, whereas the LINE port is expecting up to a volt. Input impedances are different too but that's similarly academic.

Posted by: @ldoppea

Also one important thing to note. I don't know why, but at some point during my tests, the acquisition card would stop working when pluging its alimentation to the DIY board's V+/GND connector. This was the first time this happened. Also when in this state, my computer's would alert me that the USB port was drawing too much power.

That's by design. From what I remember the standard calls for the host device (the IC that makes your USB ports work) to monitor for an  overload condition, the old USB2 standard was 500-mA, and if it detects anything it shuts down to prevent damage and informs the operating system.

Years ago, before we had such protections, that would have killed the chip and rendered the board useless.

It's not clear why a short is happening but it only takes a sliver of copper to trigger a shutdown, unlike fuses, this system works effectively instantaneously. (Yes, I know fast blow fuses exist but they are difficult to produce with a hyper accurate blow point and they have to be replaced).

Some of this might be the way you're getting your USB power which should come from the same port on the computer as you're supplying data to. The other way is for the board to be self-powered, i.e. have its own 5V supply and only connect to the computer via the two data lines D+ and D-.

This has the effect of isolating the power on the your board from the power at the computer. No "ground" is required - a genius move by the original designers which also works to eliminate interference on on the USB data connection which would otherwise cause errors. The data line is a differential pair which has the effect of being it's own ground - a technique that is widely used in electronics to isolate "grounds" in between instruments which, in very real cases might be separated by 1000s of volts! (Often it's a few millivolts but that's enough to impress noise on the signal.)

And there's the "gotcha" for the unprepared - mixing grounds can cause overload conditions destroying equipment. (An experienced repair guy, Mend it Mark on YouTube, walked into this one in recent video and shorted one of his benchtop power supplies. At least he's honest enough to admit that it was his doing even though it caught him off guard.)

Anyway, I'm not out of ideas just yet.

Let's leave the acquisition card and computer out of the picture for now and get that mic amp working.

If you have access to a *regulated* 5V supply (this is important as I'm not sure how the NMA0515 will respond if it's input section receives too much voltage and some "wall warts" (cheap ones) work on the assumption that they will be loaded that might bust the device. It probably won't but it pays to be careful. Actually the NMA0515 is similarly unregulated so it appears to produce a fairly high voltage if the outputs are measured when there's nothing to load it down (like a THAT1512 in this case).

You can get a regulated 5V directly from the USB on your computer, or better yet, a phone charger etc. Just grab an old USB charging lead, and snip the end off (the one you don't plug into the computer!).

Some, very cheap USB charge only leads only have two inner cables - red and black by convention. Most will have four though - the others are for data and can be snipped off.  The black lead goes to 0V - marked GND on your sheet and the red goes to V+ on the power connection.

You'll have soldered the NMA0515 into place so I suggest you remove the THAT1512 at first.

As we move into testing, it's important that if any of these tests fails in this order, stop immediately and get back to me as you'll have found the issue or at least, part of it.

Set your multimeter to a range suitable to measure DC up to about 20 volts (a typical range on many cheap ones these days). 

If you have a clip, you should make sure the black lead (GND, common from your meter) to any point in Matt's circuit marked as GND and we'll measure the voltages "with respect" to that point. GND really should be called "common" or "0V" but few people agree on what we should call it. There are a couple more, "earth" which goes to actual ground (literally a big copper spike stuck into the ground near your home) and "chassis" which are often mixed up but these all have a purpose.

We call 0V "common" because it's just a baseline to measure things from. We COULD use -15 or +15 as common, but that would give us different readings, so designers agree on reference point and everything is measured from that.

If you don't have a clip and are stuck with a test probe* you can hold the black probe on pin 5 (five) bottom right of the socket and measure voltages as I've noted here on your diagram.

When the THAT isn't plugged in, the NMA01515 will deliver an unregulated voltage which will be a little higher but don't worry unduly, what matters is the POLARITY - so +15v should measure 15-20 volts and the -15V should measure negative 15-20 volts (give or take).

You should get the same voltages as noted at the mic socket - but be very careful you don't short the centre pin to one of the voltage outputs as that might damage the NMA0515

If this is all OK, you can pop the THAT back into its socket (making sure it's the right way around - if I had a penny the number of times I've put a chip in the wrong way around... They usually survive but it's not good practise to try it!

The load from the THAT should tame that voltage to +/-15V and you should check this once you've established the power is being delivered correctly. If something is wrong with that chip it might (might) do something weird like drag the supply lines down.

The THAT has output overload protection (continuous) so that it will survive a short circuit and that's probably enough to drag the supply rails down *below* +/-15v. So a "quiescent" supply voltage of +15 and -15v with respect to pin 5 is a good sign.

Next check the voltage at pins 2 and 3 (again with the black lead on pin 5). There might be a few millivolts DC on those but many meters won't even register it.

IF you do sense something on those lines, stop immediately and get back to me!

Similarly, the output voltage (measured between pin 5 and 6) should be quite small or not even "move the needle". (An oscilloscope comes in handy for these tests but it's just more sensitive.)

You can leave pins 1 and 8 alone for the moment as they are connected internally to the gain of the final amplifier. When left open as we have right now, the THAT operates at a gain of x1 which means what goes in (specifically the difference in voltage between pin 2 and 3) gets spit out the other side.

You can't just connect a headphone to the output (Pins 5 and 6) to the THAT because it's rated for a minimum of 2K load. It is short protected as I've mentioned but 32R for a typical headphone is likely to be quite sufficient to trigger the "overload" protection.

You CAN (if you can get hold of one) put a crystal earpiece on this output. When I was growing up I used one even before I got my first multimeter to do all sorts of "tests" on audio circuits. (Battery powered ones!)

You can get these things from your local electronics sort or eBay. They're weird looking things that look like they belong in last century - because they do although even the Wikipedia page remarks on their continued use as "low-tech test gear" . 🙂

Crystal earpiece - Wikipedia

These curious little devices also put out a tiny voltage in the presence of sound and can be be used as a microphone although it won't work in this circuit.

image

Hopefully that's given you something to get moving with. I've just got some test boards back from JLC so finger's crossed mine will behave as I intended them to - I've put a lot of test points on mine to avoid poking around on these tiny chips but nothing is guaranteed to work until it does.

Marc


 * Like most of us who only do this occasionally and haven't invested in some decent test gear (that includes me, but don't tell everyone).


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 23/02/2024 3:12 pm
(@ldoppea)
Posts: 30
Eminent Member
 

Hi,

 

New weekend = new batch of tests 🙂 

 

Posted by: @marcdraco

Oh, I wish other people would record their tests as well as you - makes it so much easier to figure out what's going on. Don't lose faith, you're doing well!

I'm a developer so I know the pain of debuging without correct logs 😋 

 

Posted by: @marcdraco

When you use the word "alimentation" do you mean "power"? I'll assume that's what you mean given that it seems to be in context. I'll use power in my replies for the native speakers who might get tripped over that.

Woups sorry, yes I meant "power", "alimentation" is the french word for "power"

 

Posted by: @marcdraco

Matt's board is specifically designed to drive the green (LINE) input on these devices.

I'm not sure to get this. In the acquisition card, the green input is documented as an "output only" line 🧐 

image

Posted by: @marcdraco

making sure it's the right way around - if I had a penny the number of times I've put a chip in the wrong way around...

I plugged it with the chip's "half hole" thing aligned with the one on the diagram. Is that correct?

image

 

Posted by: @marcdraco

You can get a regulated 5V directly from the USB on your computer, or better yet, a phone charger etc. Just grab an old USB charging lead, and snip the end off (the one you don't plug into the computer!).

I took the one from my computer as it was easier (regarding cables length). My voltmeter measures 5.12V on it

 

Here are the measurements without the THAT:

image

 

Here are the measurements with the THAT plugged:

image

Those are different from what you expected, so I stopped my tests here. Hope that will help

This post was modified 8 months ago 2 times by Ldoppea
 
Posted : 02/03/2024 3:03 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

A developer, that explains it! This is making it a LOT easier to see what's going on. (Are you French or French-Canadian perhaps. I know there are other places around the world where French is the spoken language of course. I'm just curious, your English is excellent but I can detect it's not necessarily your first language. I envy that amazing ability.)

Anyway, this is fascinating.

You everything aligned correctly which helps of course. 

Those voltages speak to something very wrong though. The voltages with the THAT unplugged show your power supplies are working perfectly and are correctly wired and the accuracy reflects the voltage "drops" around the circuit. The tiny voltage differences of a few hundredths of a volt are caused by the tiny resistances in the strips of copper, solder joints and meter accuracy.

So you can see that the unregulated 15v and -15v "rails" are working as they should, I'm impressed that the NMA0515 manages to to get so close, although that's largely a function of a good design assuming a well regulated 5V supply on the way in.

Your measurements with the THAT plugged in are very telling though, this is an excellent demonstration of how even a simple instrument can be invaluable.

My gut reaction is you may have a faulty or even a "fake" THAT, perhaps a performant single op amp (like an NE5534) re-printed to appear to be much more expensive chip. 

Check the pinouts and you'll see they are surprisingly similar because the power supplies (pins 4 & 7 with 5 as a 0v/gnd), inputs (2 and 3) and output (6) have the same function on the face of it.

Now you'll notice that the voltages drop a different amount and also that there is something off with input pin 3. Both pin 2 and 3 are inputs they go to a bipolar transistor so they *should* sit at around 0v. There's a fiddlingly small amount of leakage that appears across the input resistors - 3.9K in Matt's design - are in nanovolts so it's not worthy of discussion since it's removed as a function of how a differential amplifier works).

Looking back to the NE5534 single amp, you'll note that pins 1 and 8 have a different function - which is to correct any DC output error when the circuit is operating. It's only used when manual calibration is required. You'll note they serve the "gain" function in the THAT1512. Pin 5 called "ref" on the THAT and "compensation" on the 5534 is the reference voltage, ground for our purposes.

Pin 8 doubles as compensation pin - this is where we put a small capacitor to "compensate" for the phase shift caused by the amplifier so it doesn't go into free-running oscillation. A full discussion of this isn't relevant here but it's worth a read if the inner workings of undercompensated amplifiers are your bag. For most purposes, an internally compensated amplifier, which comprise the majority of amps in the wild is far more convenient. The NE5532 is a dual version of this design, favoured by audio engineers because it gives all the great features of the 5534 but the internal compensation does away with the need for an external capacitor and since most audio circuits are AC coupled the small amount of DC offset (usually a few microvolts of error) that appears at the output is just harmlessly blocked.

Given that, consider the measured voltage on 3 and 4 is quite telling. It's almost as if something is partially shorting those two pins - but that doesn't happen in a vacuum.

If you had a short near the chip, I'd expect the voltage to show up even without the THAT plugged in. The fact that it only appears when it is, is a concern and is clearly a symptom of what's going on.

-3.6 V shouldn't appear on the input - at all!

We can ignore the output voltage to some degree as that's a function of the input - but again here, the difference between pins 2 and 3 is -3.6 V and with a gain of 1 (pins 1 and 8 open) the output would be -3.6 V too. At least that's what it would be in a circuit that, for whatever reason, applied -3.6 V to the input pin by design.

There's quite notable voltage sag the on the negative supply which suggests that there's a load pulling that down hard*.

So the output appears to be an amplifier running flat out - the output is "saturated" (clamped as hard as the chip is able to go).

If you look at the specs for the chip, you'll see it is capable of a 13.5 volt swing on 15 volt supply. Which means the output circuit is stops at about 1.5 volts short of each rail (it'll be a very carefully designed class AB power stage).

But this when the amplifier is actually "amplifying" something and it's supposed to be outputting -3.6 volts.

Just in case I've explained that badly, consider that a class AB amplifier is limited the voltage at it's power supply (this is true of power amplifiers, but circuits exist to boost beyond the input, as seen in the NMA0515).

So the theoretical maximum output swing for a bipolar power stage is about 1.5 volts shy of either supply. MOSFETs and specially designed "rail to rail outputs" subvert this problem, but this is output is about quality so no fancy tricks here.

Your measurements prove this effect brilliantly (sadly). If the output is saturated because the amplifier is flat out, one of the supply lines will probably sag a little, and look at that.

Pin 4 (Vee) sits at -12 volts DC.

Pin 6 (Output) sits at -10.5 volts DC

The difference of -1.5 volts (give or take a few millivolts) is immediately indicative that the output is saturated to the negative supply.

This is worrying.

To explain, the guts of the THAT are essentially three differential amplifiers configured as an instrumentation amplifier. (My Michelle board does exactly this with discrete amplifiers for very low cost at the cost of quality [the SMD version anyway].)

Thinking of this as a black box, recall that it's supposed to configured for a gain of 1, and the output is saturated, the gain is clearly running amok!

Normal operational amplifiers require negative feedback to control their gain (usually a network of two resistors). Without that control their gain is in the millions or even billions. Indeed, this is the essence of Harald Black's discovery of negative feedback.

This is a very deep and fascinating discussion that's definitely worth a read. 

In essence Black discovered that if he made an amplifier stage with a theoretical mathematical gain of (say) one million and then threw away almost all of that gain, which sounds barmy, the following happens compared with an amp designed for a gain of (say) 10:

The input impedance is multiplied massively.

The output impedance is reduced into fractions of an ohm.

Noise is reduced.

THD is reduced greatly, creating an amplifier that's not just predictable but also performs with extreme linearity.

That last one one made my brain squeeze out of my ears the first time I heard it. But (another aside) if you look up Crossover Distortion at your favourite site you'll see it's a really nasty issue in Class B output stages.

So much so that it's a deal-breaker for many (some experts posit this may explain the idea that valves are better than transistors because of this transistor sound.

You'll see from the pictures (I won't repeat them here) that for a short period, when the signal nears the 0V rail, both transistors are off. This adds a (from memory) a second harmonic blip which is very audible.

Most audio amps these days use class AB bipolar or, more commonly, class D output stages which are efficient but only realisable easily with MOSFETS. Operational amplifiers are produced in the millions for any number of uses and class AB output stages are power hungry. (The NE5532 is a great example. Check out the quiescent [standby] current per amplifier of the 5532 (about 4-5 milliamps) vs. the class B powered LM324: 300 microamps! Later designs have improved on the 5532 greatly but they cost more naturally.

Such a distortion would seem to be a deal breaker in an amplifier that might be inside a sensitive instrument but negative feedback reduces that distortion to levels that are effectively inaudible.

TL;DR

That -3.6 volts is coming from somewhere, but it's fairly clear the chip is operating at maximum gain (10,000). Since that works out at 36,000 volts (!) the output just goes as far as it can - which is about 1.5 volts short of the rail: exactly what we're seeing. The rail itself is being pulled a little short of its expected -15 volts because the amp is being driven so hard.

Which makes me believe the THAT has gone to silicone heaven - perhaps from ESD? Any voltage appearing on an the input pin of a differential amplifier is a sign something is badly wrong. When ESD fries a chip it's difficult to predict what the damage will do, but it appears to be in the gain selection stage. 

Take a look at Figure 1 in the THAT's datasheet. RG1 and RG2 are connected to the emitters of the input differential pair (this is possible because the whole design is on a single chip) and they are controlled by separate current sinks (that's the overlapping circle symbol).

Current sinks work by "adjusting" the voltage at a position in the circuit to keep another point in a circuit drawing a fixed current. We don't need to know actual current but it's probably two current mirror circuits and this is where I think it's shorted the base-emitter junction of the transistor on the non-inverting side.

That -3.6 volts is probably supposed to be setting the designer's quiescent point on the emitter of that transistor and since it's shorted to the base internally what you're reading is exactly that. I'll be lazy and assume that Rb has gone short (they'll almost certainly be in the same part of the die) which might explain the high gain.

Not the news I'd wanted to deliver. It's entirely possible this was ESD but with suppliers these days it's also possible (and with eBay, probable) that the chip was DOA.

If you would be so kind as to do the voltages at other points in your circuit with the THAT in and out (don't worry about making it pretty) I can hopefully see if anything else is out of whack.

You should probably contact your supplier and ask for a replacement under warranty, it's worth a try.

Marc

 

 


*During my first attempt at improving Matt's design, I didn't check the output specs on the NMA0515 and figured it would drive several NE5532s. Short answer: it won't. Longer answer, I know better, I should have read the spec sheet and I deserved to get my ass spanked for such a basic error. Longer answer still, well more complicated, it's doesn't take a genius to figure out that increasing voltage by a factor of 6 is going to cost something... ugh. Go to the back of the class Marc and play with some blunt crayons.

 

I think it's better to admit to these mistakes, esp. as an experienced person, because when someone seems able to take this stuff in their stride, we assume they don't make errors. But we do. A lot. 😉

 


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 02/03/2024 8:30 pm
(@ldoppea)
Posts: 30
Eminent Member
 

Posted by: @marcdraco

Are you French or French-Canadian perhaps. I know there are other places around the world where French is the spoken language of course. I'm just curious, your English is excellent but I can detect it's not necessarily your first language. I envy that amazing ability

Thanks for the compliment. Unfortunately I'm way less confident when speaking english that when writing it (and also when the topic is not a tech one). I'm from France, in the Alps.

 

Those are kind of bad news, but this is consistent with my first intuition, based on the fact that there were multiple people complaining about fake THAT1512 or fake NMA0515, all bought on Ebay or Aliexpress.

 

I'll try to negotiate for a free new one, but i'm afraid this will be too late as I bought it 3month ago. If I need to buy a new one, the choice would be to try Aliexpress again, or to secure it by buying it on a professional electronic shop, but the one I know will charge 23€ just for the shipping 😨 

 

Posted by: @marcdraco

If you would be so kind as to do the voltages at other points in your circuit with the THAT in and out (don't worry about making it pretty) I can hopefully see if anything else is out of whack.

 

Here are the results. I tried to be as exhaustive as possible, even if measurements were on the same line, so we can be sure there are no bad contact.

I did not measure contacts under some capacitors as it was difficult to access them from the top of the card, and when I try to measure them from the bottom side I tend to loose track of what I'm measuring.

Also some measurements near the capacitors may be wrong, I got 2 or 3 points where the Voltage was decreasing slowly when measuring. So if something is abnormal, tell me and i'll measure it again.

THAT unplugged:

without that

THAT plugged:

with that
 
Posted : 03/03/2024 3:59 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

Gosh darnit, I thought I'd replied but I got distracted and then Windows decided to restart at random and ...

Anyway, at least we know it's not a "fake" but that doesn't mean it arrived working either. ESD is the likely culprit somewhere along the line though. If the air where you live is dry (usually cold air is very dry and we've had a dry winter, ESD is going to be worse problem. It's much less so during the summer months because the air holds more water and that seems to make it difficult for static charges to form. Living where you do has my "ESD Spidey Sense" tingling, or maybe the cat has been laid on my arm for too long?

Plugging it in the wrong way around or applying the wrong voltages isn't going to cause that, the inputs are protected by diode limiters but even though they do help with ESD (static discharges) it's all to easy to get caught, particularly if the air is crisp, you're wearing certain fabrics and walking on a carpeted floor. It's not necessary to build up enough charge so as to get a "zap" on a door handle (that amount can kill a discrete bipolar transistor in some cases), the die is very fragile the near instantaneous rise in temperature can either blow a hole through, cause random shorts or both. CMOS (including MOSFETs) are especially problematic in this regard.

Once the IC is in circuit, ESD is less of a problem as the charge invariably leaks away harmlessly to ground before it can cause any problems. Invariably: never say never.

The "correct" way to do this is to use earthed work mats and earthing straps but none of us got time for that merde am I right?

The solution is to touch something like a central heating pipe (any cold water faucet works too) and that will bleed off any charge you're carrying. Ideally, chip insertion is the last part of the job so you can take the board (socket soldered in place ready) as near to your ground point as possible and ground yourself before you remove it from it's protective carrier tube/foam.

That postage in France seems rather high... Is it really that much?


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 03/03/2024 11:38 pm
(@ldoppea)
Posts: 30
Eminent Member
 

Hi,

Posted by: @marcdraco

The "correct" way to do this is to use earthed work mats and earthing straps but none of us got time for that merde am I right?

I don't have one, but when doing electronic stuff I always try to discharge myself by touching the back of my computer case (this is an old habit I got reading computer manuals when I was younger). But I suppose this is not enough?

But you are right, inside air is pretty dry due to heaters as it is the winter season. Fortunately carpeted floor are nearly inexistant in France so the risk is not too big.

 

Posted by: @marcdraco

That postage in France seems rather high... Is it really that much?

When talking about this kind of electronic, yes... Seems like consumer shop specialized in electronic in my city did not survive the ebay/amazon era, the last one I new has closed not long after Covid.

So today the only French option I know is to buy on retails that target companies. And because they target companies, their cheapest shipping option are like "express with signature" ones.

I contacted the Aliexpress seller, but without any surprise, they didn't even answer my message.

Anyway, I may have found a solution to get a chip from one of my parent's friends. Maybe next week. 

 
Posted : 07/03/2024 11:40 pm
marcdraco reacted
(@justmike12)
Posts: 4
New Member
 

@marcdraco Hello, I am new to the forum and pretty much a beginner with electronics.  It seems you have contributed quite a bit to the forum.  I am looking to build this microphone and noticed this particular post when scrolling through and wondered if you could help me.  It seems some people have had a few issues with build, and from your post I could focus on 1) Replacing the 2u2 Electroytic Capacitors with Polyester Film to reduce noise; 2) Not use /or be very careful with Strip Board as it is very easy to short the components through "solder jumps" (Perhaps I could get a PCB printed to help reduce this issue?); 3) Add a RF snubber in there.  Did you mean RC Snubber (When I researched what a snubber was- (and my understanding is it is a device that deals with sudden voltage spikes - which cause electromagnetic interference - when there is an interruption in the flow of current) one of the options was RC snubber).  Would this be a resistor and a capacitor, and where would they go in the circuit?

 

I assume Matt's V2 isn't out yet?

 

I think that's it for now, thanks in advance!

 

 
Posted : 08/03/2024 12:29 pm
(@taran)
Posts: 5
Active Member
 

I also just wanted to check if there was any update on the V2, I've been thinking about building this for close to a year now but didn't want to start if it's just around the corner

 
Posted : 12/03/2024 5:04 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

"Just" around the corner? Well it *is* strictly speaking waiting on some (I hope final) adjustments which I'd planned for today. This is entirely on me. It'll confess it's pretty soul crushing dropping a bunch of your own money only to find that one (one!) dumb error which I missed due to familiarity breeding contempt brings the whole thing to its knees. 

I've split it into several smaller parts which have developed out of the need/demand for greater versatility. Such is dev. work and with on average a 6-8 week turn around on development versions it's stretched me further than I'd imagined in a quest for better specs.

A side-project loosely based on the Jensen 990C discrete op amp worked out of the gate and performed better than I'd imagined being capable of driving a pair of over-ear headphones.

I've found that my own PCs generate an interminable amount of noise on the USB C which has had me chasing my own tail looking for a solution. I'm still not happy with the amount of bleed though, so this may result in an externally powered design. (Both are possible with the board and some external bits/bobs.)

TL;DR

I've included some photos of the current state, so you can see these aren't vapourware.

V2 mainboard: I slung a classic LM3916 VU meter on there because it seemed like a nice bit of polish and there was just enough room left. This is entirely optional and easily left off. The rest of the board is a six amp discrete version of the THAT1512 with a much simpler volume adjustment and a real-time headphone amp (recording only, no USB playback!) The main selling point of this one is it works at 5v and still manages quite respectable performance. 

V2 head units. Several of my older designs work well with this but I've slowly pushed outward to support a more diverse collection of capsules, specifically the 34mm range of professional, unbiased capsules. This is the first prototype but I figure I need to put some solder jumper options to make things easier.  

Both are four layer boards with excellent RFI rejection.

image
image

That "version" of the project has fairly demanding power supply requirements 48 volts split as two 24v rails. But it's currently nowhere near to being ready. Ironically, the poor-man's Jensen 990 is so good I'm tempted to lay out something more convenient for this and use it in instead. Here's a shot of the (non working/untested) prototype. If you look really close you'll see I cheaped out on the input capacitors but I can't test this until I work out a safe way to get a decent power supply. 

As usual there's a poor-man's way around this one but it's a bit of a chew on so this board (which is designed to drive professional headphones at HiFi quality) will have to wait. It's mostly 5532s (or better as they are all socketed, with a Sziklai class AB output and a fully buffered balanced output.  

image

All my schematics are will be Open Source naturally, but I'm holding back until I've got 100% tested prototypes.

My cheaped-out Jensen 990 looks like this - I'm sorry the picture is so poor, I need to figure out how to plot these for everyone 😉

image

The main difference is this has an optional matched "pair" baked in and an adjustable current mirror to allow for the different configuration. And I've dropped much of the frequency correction, most notably the inductors and balancing in resistance in the emitter circuit. These components were specific to the, long discontinued, LM394. The capacitors on this board need to be NP0 or some poly- type vs. the more usual ferrited-MLCCs so I've left space for through hole ones if required. Even as supplied with the "off-the-peg" transistors, the performance is pretty impressive. 

If you can follow a schematic, this can be made up on stripboard, but there are some tricky bits which aren't really practical on strip. 

I'll admit this one gives me chills because it's relatively inexpensive in volume and only needs a little more TLC to improve things like output drive current by adding more or larger output transistors. Getting to play with my own versions of legendary circuits from my childhood is a like drug. Which also means it slows the rest of it down since I only have a limited amount of time. This, again is entirely on me.

This post was modified 8 months ago by marcdraco


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 13/03/2024 11:05 am
(@justmike12)
Posts: 4
New Member
 

Any thoughts on this when you have a chance?

 
Posted : 13/03/2024 6:12 pm
marcdraco
(@marcdraco)
Posts: 581
Moderator
 

@justmike12 Hi Mike and welcome to our little corner of the web. 

As things are today, I've (just, as of last night) sent a rev. 3 of the Varee head over to JLCPCB. A couple of gremlins crept in as happens with prototypes. I've got it on a rush so I should be know if it's all polished up and works without having to solder bits all over the place. (If I had a machine to remove SMD parts, I'd make a fortune selling used parts.) On paper the performance is almost off the chart - but I'll wait and see so I don't look like an idiot.

So that's the top, but it designed to work with 48 v phantom and should work all the way down to 5 v but that's something I need to test against the revised "main" board. Those came back due to a fault (mine) and since I took my eye off the ball, I slipped up on that too. So that's another couple of weeks for me to sort those issues but it does work at 5 v and it's lower cost I suspect since I've removed the voltage doubler and THAT1512. 

I  can only apologise this has taken so long (for all of you watching). I did get a case of "just-another-feature-ism" for too long, so I kicked that to the kerb and went back to basics and produced a professionally specced PCB that fits capsules of 25 and 34mm. 

The board above (Garrie) came back with that order but I'm still figuring out a practical power supply that is safe for everyone to build and time isn't on my side. The 5 v one (Michelle) is probably going to be enough for most people; Garrie is a few volts short of a fully professional headphone driver, so I might go the whole hog on that.

The experimental Jenson 990 (I'm naughty calling it that because it bears little resemblance to the original) has proven surprisingly good and opened the possibility of expansion to a board that (in essence) reproduces the same functionality as the THAT1512. I have that on a back burner for now.

I have a few boards left over that I'll auction when I finalise the design and perhaps change the, shape of the board. It drops into any circuit where a bipolar operational amplifier is required and it matches and usually exceeds, the performance of the NE5532 especially driving capacitive loads. Swapping out the input stage for the MAT12 blows away pretty except the most expensive amps and it's way cheaper even with the MAT12 (about £8-£12 per package). This is something that only came about as part of this project, so I can only give a H/T to @diyperks for that.


Take everything I say with a pinch of salt, I might be wrong and it's expensive way to learn!

 
Posted : 15/03/2024 5:54 pm
xenon
(@xenon)
Posts: 10
Active Member
 

Hi everybody, it's my first time here.

Some days ago I saw the video and was captivated by the idea, I would like to replicate the project but I first have a question.

I was wondering if I could achieve a similar or better result using the JLI-3412 instead of the JLI-2555.

image 2024 03 17 144239727

It's a little bit bigger but has a better Signal-to-Noise ratio and a similar sensitivity. I am an electronic apprentice but my knowledge in the audio field isn't very much, and I am worried that the JLI-3412 may need a different transistor or something.

Thanks in advance.

This post was modified 8 months ago 2 times by xenon
 
Posted : 17/03/2024 1:43 pm
(@justmike12)
Posts: 4
New Member
 

Hello, thank you for your warm welcome, and thanks for the post.  I'll be keeping an eye on your progress!  Sorry to have confused matters but this was the post I was referring to to see if you had seen it:

"Hello, I am new to the forum and pretty much a beginner with electronics.  It seems you have contributed quite a bit to the forum.  I am looking to build this microphone and noticed this particular post when scrolling through and wondered if you could help me.  It seems some people have had a few issues with build, and from your post I could focus on 1) Replacing the 2u2 Electroytic Capacitors with Polyester Film to reduce noise; 2) Not use /or be very careful with Strip Board as it is very easy to short the components through "solder jumps" (Perhaps I could get a PCB printed to help reduce this issue?); 3) Add a RF snubber in there.  Did you mean RC Snubber (When I researched what a snubber was- (and my understanding is it is a device that deals with sudden voltage spikes - which cause electromagnetic interference - when there is an interruption in the flow of current) one of the options was RC snubber).  Would this be a resistor and a capacitor, and where would they go in the circuit?

 

I assume Matt's V2 isn't out yet?

 

I think that's it for now, thanks in advance!"

Look forward to hearing from you 

 
Posted : 17/03/2024 6:17 pm
Page 21 / 28