DIY preamp for dynamic microphones

[Aranwe]

Hey everyone, I just wanted to share this project I've been working on.

Some of you may have seen DIY Perks's video where he makes a DIY USB microphone. I really liked the idea so I decided to build one myself. Mechanically his design is amazing, but I wanted to make some improvements on the electronics side  

 

The preamp used in the original video was designed to work with electret microphones. These are a special type of condenser microphones that require a very low voltage opposed to true condenser microphones, that require 24 or 48V phantom power. Electret microphones are therefore commonly used in portable devices and are often very small, which is why most people associate electret microphones with poor quality. There are a few capsules with bigger diaphragms like the JLI-2555BXZ3-GP, but those can be hard to find.
My preamp is a multi stage transistor based preamp that offers high gain, so it can work with dynamic microphones. It is built around the BC848 NPN transistor and is designed to boost a microphone signal to a suitable level for an ADC converter. 

 

1. Stages

There are 4 stages in total.

• The input stage is a common emitter design that offers a high input impedance and adds +18dB to the signal.
• Next are two low harmonic distorsion differential amplifiers that boost the signal to the desired output level.
• Finally there is a common collector circuit that offers a low output impedance.

On the PCB there are multiple jumpers to uncouple the stages and probepoints to test each stage individually. This also allows to bypass one of the differential stages if less gain is required, or add another preamp if you need more !

 

2. How to build one
If you want to build a preamp for yourself, all the design files and specifications are on github (https://github.com/AranweLTT/mic_preamp). I am still actively working on documenting the project and improving the design, so any feedback is welcome ! Please let me know if anybody would be interested in a kit, or even ready to go PCBs available for purchase.

3. Does it actually work ??
I ordered a first prototype on JLCPCB and surprisingly enough there was only one major issue : I forgot a trace in the current mirror for the differential amplifiers so I had to fix this. Other than that I measured the gain above +46 dB, and tested it out with a dynamic microphone which worked flawlessly ! Even though the preamp has a micro usb input for power, I used a lab power supply because even after filtering the power is very noisy over usb… 

 

3. Improvements !

I'm already working on a amplified headset output so you can monitor the output directly (software monitoring has an aweful delay, so you either go hardware monitoring or you go home). I'm also planning on integrating a codec directly on the pcb rather than having to add one downstream. This way it is a true USB microphone (for now the micro usb port is only for power!! No data !!)

 

 

Please let me know if you have any ideas. Also if there's any 3D designers that can help me out with designing an enclosure your help would be more than welcome !)

Tschüss !

 

 

 

 

 

schematic_preamp_v4_2.pdf

[H713]

One concern I have with this is that you don't have balanced inputs - that's a problem for microphone signals. You might get away with it if the cable is short, but balanced inputs on a microphone preamp are important for keeping hum and noise down.

 

You also seem to be using a lot 1 uF caps in the audio path, presumably those are the MLCC ceramic caps- careful. Ceramic caps are often microphonic, and they also tend to be rather nonlinear. Unless they're C0G/NP0, keep them out of the audio path. They're good for bypass caps though.

 

All that said, it's nice to see someone who isn't afraid to design with discrete transistors.

 

 

[Aranwe]

Thanks a lot for your feedback !

11 hours ago, H713 said:

One concern I have with this is that you don't have balanced inputs

I agree that a differential first stage would have been better, but I didn't find a way to get a high enough input impedence. That's why I went for a common emitter, because a bypassed differential stage has an input impedance of Rb (base resistor) that can't be more then 10k or it will mess with the polarisation (bias). Please let me know if you have any ideas !

 

Edit: Designing with a 5V power supply was very challenging, but it allows the preamp to be bus powered. A symmetrical 15V supply is often used in audio gear because it makes it easier as you don't have to deal with any biasing. Also for the same current I can take bigger resistors so higher input impedance across the board. Maybe I can use a step up converter but a switching power supply on a senstive preamp doesn't seem like a good idea !

[H713]

10 hours ago, Aranwe said:

Thanks a lot for your feedback !

I agree that a differential first stage would have been better, but I didn't find a way to get a high enough input impedence. That's why I went for a common emitter, because a bypassed differential stage has an input impedance of Rb (base resistor) that can't be more then 10k or it will mess with the polarisation (bias). Please let me know if you have any ideas !

 

Edit: Designing with a 5V power supply was very challenging, but it allows the preamp to be bus powered. A symmetrical 15V supply is often used in audio gear because it makes it easier as you don't have to deal with any biasing. Also for the same current I can take bigger resistors so higher input impedance across the board. Maybe I can use a step up converter but a switching power supply on a senstive preamp doesn't seem like a good idea !

Nothing inherently wrong with using a switching converter on a mic preamp - I've done it, and it works pretty well. Companies like Meanwell makes some nice little potted DC-DC converter modules. You can get a dual output, isolated 5 V to 15 V converter for something like $6. They're reasonably quiet, but if you're really worried about it you could add some linear regulators or capacitance multipliers. That said, your 5 V USB power source (whatever that is) is likely to be quite noisy. 

 

Regarding balanced inputs; microphone preamplifiers have always been a significant design challenge. For decades, you really couldn't get a decent microphone preamplifier without using an input transformer. This transformed the relatively low output impedance of the microphone into something more optimal for a typical BJT amplifier, and it typically gave a useful 10 - 20 dB of gain. While these transformers aren't ideal by any means, they didn't sound bad (and a lot of people believe their distortion is a good thing), and the really good ones made by Lundahl and Jensen were really pretty linear. The big problem with them is cost - a good microphone transformer ranges from $50 - $250 each.

 

Later, as op-amps got better and engineers got smarter, you started to see most microphone preamplifiers using a mixture of op-amps and BJTs, and the topology used in the Soundcraft 600 became pretty common. It performed reasonably well, and it saved the size and cost of an input transformer. A more refined (and lower noise) version can be seen in the Crest V12 console, the schematic for which I've also attached. As a bonus, the Crest V12 is a much newer desk, and that means legible schematics.

 

At some point in the 1990s, Crest Audio (well-known at the time for their 8001 power amplifier) bought the company of a man named Jim Gamble. As a result, Crest consoles showed a strong resemblance to the legendary Gamble EX56.

 

Also, here's another good reference on these topologies:

http://www.thatcorp.com/datashts/AES129_Designing_Mic_Preamps.pdf

 

 

 

 

 

[H713]

Also, because this may serve as a useful reference, this is the thread I started over at DiyAudio when I designed my first discrete microphone preamplifier. This was also the first discrete transistor circuit I designed, so I made some dumb mistakes. Nevertheless, there's some good information to be gleaned from that thread, and the final circuit actually performs quite well. 

 

https://www.diyaudio.com/community/threads/solid-state-balanced-microphone-preamplifier.337494/page-2

 

As a note, that was also the first PCB layout I ever did, and consequently probably the crappiest layout I ever did. It worked okay, but if you want to use that circuit I strongly suggest redoing the layout. It works as it is, but the lack of a ground plane and lazy decoupling is pretty obvious when looking at a step response.

 

I keep meaning to come back to that project and clean it up - new layout, recalculate a few part values, but just haven't had time. 

 

[Aranwe]

That complementary feedback structure looks really promissing ! I'll have to re calculate the values though, and simulate it

 

[Aranwe]

Here's a VERY EARLY look at the next version I'm working on. Compared to the previous version I added

• a Class AB amp (stereo)
• integrated USB Codec
• Gain control potentiometer
• 6.5mm input jack instead of 3.5mm
• Proper mounting holes

Schematic_audio_interface_2022-03-27.pdf

[mariushm]

Well, I like it, but there could be some improvements in the layout.  There's lots of small things that jump to me. 

Pretty much you could rotate all the 3 pin transistors 90 degrees to the left, basically like the one on the top left corner and you'd get better layout. 

I'd also rotate the HDR1 jumper header 180 degree and move the printed text HDR1 to the left of the header , just like the INPUT 2 header.  Then, move the HDR2 header to be in line with the HDR1 header 

I'd also place that header by the headphone jack in the unused space behind the headphone jack  and basically be consistent, have all headers in the same orientation if possible. Why have some horizontal and some vertical? 

Don't use aluminum electrolytic surface mount capacitors... use through hole, they're more stable long term.

No idea why you use individual BC848 when you could use pairs for example MBT3904 : https://www.digikey.com/en/products/detail/onsemi/MBT3904DW1T3G/1477283

 

For at least a couple of your transistors you could get pre-biased transistors and save room on the pcb by not having to add the 10k resistors : https://www.digikey.com/en/products/detail/nexperia-usa-inc/PEMH9-115/1157407

 

I suppose you'd want to use as many components as JLPCB has in their inventory... 

 

for example : the top right corner (with the VS4 pad) could be arranged like this and get the same result  :