Small mic pre/mixer - now with pics!

I need a small, portable mixer with 8-10 mic inputs. Most small mixers I have seen have only 2-4 mic inputs - are there any (cheap) exceptions?

So I thought about a DIY solution. This is what I have come up with so far:



A standard mic pre circuit followed by a pan pot. The outputs should then go into the inverting inputs of two op-amps. The header on top is for an "upper deck" board with a phantom switch and perhaps a peak detector. I plan to use small boards that can just be linked together in a chain (to fit in a 1U rack or other small box).

Does this sound like a workable solution?

Best regards,

Mikkel C. Simonsen

Do you regulate gain by changing phantom power voltage?

I think that's an eminently reasonable solution. A few suggestions:

1) I'd make R4 = 1k82 ranther than 2k7. That'll wind up with a load on the microphone close to 1k5, which is what a lot of microphones seem to like. If you want to get fancy, you could add a 750R resistor to be switched in parallel with R4 to give you an input impedance of about 500 ohms, which microphones like the SM57 would like.

2) Any chance you could make L1 and L2 into a single common-mode inductor?

3) Robert Orban, who designed that panpot circuit (more or less), suggested making the resistors surrounding the pot 0.707 x the pot's value. The nearest 1% value is 7k15. [Edit: Or did he suggest 1.414 x the pot's value? Right now I'm too sleepy to remember. If it's the latter number, then 14k3 is what you want.]

One other thing to keep an eye on: if you're using an inverting opamp as the summing amplifier, make sure something else inverts the signals' polarities someplace in the chain. Either another inverting stage, or hook up the inputs in reverse polarity.

Peace,
Paul

I haven't used the 1512 yet
or done any tests

but gut feeling tells me to use an opamp to buffer to the pan and sum section

the 1510/12 may be better on it's output than the 2017

[quote author="pstamler"]I think that's an eminently reasonable solution. A few suggestions:

1) I'd make R4 = 1k82 ranther than 2k7. That'll wind up with a load on the microphone close to 1k5, which is what a lot of microphones seem to like. If you want to get fancy, you could add a 750R resistor to be switched in parallel with R4 to give you an input impedance of about 500 ohms, which microphones like the SM57 would like.[/quote]
This will be used mostly with Audio Technica instrument mics (electrets) - I'm not sure what they like. I just went with the normal 2k recommendation.

2) Any chance you could make L1 and L2 into a single common-mode inductor?

Click to expand...

Yes, I haven't finished the PCB layout yet. Any value/type recommendations?

3) Robert Orban, who designed that panpot circuit (more or less), suggested making the resistors surrounding the pot 0.707 x the pot's value. The nearest 1% value is 7k15. [Edit: Or did he suggest 1.414 x the pot's value? Right now I'm too sleepy to remember. If it's the latter number, then 14k3 is what you want.]

Click to expand...

The values I have used were "stolen" from a Rod Elliot schematic. Anybody else remember if 0.7 or 1.4x is the optimal value :grin:

One other thing to keep an eye on: if you're using an inverting opamp as the summing amplifier, make sure something else inverts the signals' polarities someplace in the chain. Either another inverting stage, or hook up the inputs in reverse polarity.

Click to expand...

Yes, I just planned to reverse the connections on the output.

Best regards,

Mikkel C. Simonsen

[quote author="Kev"]I haven't used the 1512 yet or done any tests[/quote]
Get on with it then! :wink:

but gut feeling tells me to use an opamp to buffer to the pan and sum section

the 1510/12 may be better on it's output than the 2017

Click to expand...

Yes, I remember reading this about the 2017s - on one of your webpages I think. The That datasheet just mentions that the min. load is 2k - not if the 1512 is comfortable with a 2k load or not... The min. load in this case would be 2k5 - or more if Paul is right about the pan circuit. Has anybody else tested this with the That chips? I have only used them with a 50-100k load myself so far...

Best regards,

Mikkel C. Simonsen

your not shown peak detector may decrease min.load further, so buffering is not a bad idea. As inverting buffer you also avoid phase turn in your summing amp.
I would miss a pad at the input section cause your actual mics could change. A phase reverse could be supplied by an xlr-adapter if required.

2) Any chance you could make L1 and L2 into a single common-mode inductor?

Click to expand...

Yes, I haven't finished the PCB layout yet. Any value/type recommendations?

Click to expand...

I've simulated and experienced a lot on transformerless mic preamp RFI protection and I must admit that it is a tricky business. A simple LC filter as shown on your schematic is very dangerous. Some microphones (e.g. the modern Schoeps variety) has very low (and presumably inductive) output impedance which will give you a very high Q, resulting in monstrous peaking (15 dB and more). This can even lead to instability problems as the output will feed back to the input at high gains. Solutions include a resistor (about 100 ohm) in parallel with the inductor or a snubber network (see the 9k schematic).

Nonetheless I'd prefer the suggested CM choke. I'd go with about 50 uH, add 100 ohm resistors in parallel (again to control peaking) and use two 1 nF capacitors to chassis afterwards. Farnell 926-5783 or 926-5791 look like suitable parts for the CM choke.

Regarding loading: you could try a discrete buffer inside the output stage loop if you don't want another opamp.

Samuel

[quote author="mcs"]
The values I have used were "stolen" from a Rod Elliot schematic. Anybody else remember if 0.7 or 1.4x is the optimal value :grin:

Best regards,

Mikkel C. Simonsen[/quote]

I typically target -3dB change between pan pot centered and hard panned L or R. The values shown only change about 1.5 dB. There are probably multiple combinations of values that meet that criteria but 4x 13k resistors around a 10K (linear taper) pot works if my early morning calculations are accurate. I also show about 10 dB of loss so you need 10 dB of makeup gain in following stage.

JR

PS to maintain proper polarity from input to output you can just swap +/- inputs for odd number of inversions. In a console you also want to be sure all inserts and such are in proper polarity wrt inputs.

edit: 10dB makeup gain is wrt pan pot so 43K resistor is close.

I sent you an email Mikkel,

cheers,
Ruairi

Notes On Panpots, Robert Orban, JAES, Dec. 1971, Vol. 19, Number 11.

I'd post it, but I don't want to step on any toes with regard to copyright.

[quote author="Harpo"]I would miss a pad at the input section cause your actual mics could change.[/quote]
I don't think a pad will be nescessary. For this application I have always needed 30-50dB gain with different mics. I should be able to get gains as low as 0dB with the That chip.

[quote author="Samuel Groner"]I've simulated and experienced a lot on transformerless mic preamp RFI protection and I must admit that it is a tricky business. A simple LC filter as shown on your schematic is very dangerous. Some microphones (e.g. the modern Schoeps variety) has very low (and presumably inductive) output impedance which will give you a very high Q, resulting in monstrous peaking (15 dB and more). [/quote]
So far I have only used a pair of ferrite beads, so I don't think they do much harm (or have much effect for that matter). The peak would be at a very high frequency at least.

This can even lead to instability problems as the output will feed back to the input at high gains. Solutions include a resistor (about 100 ohm) in parallel with the inductor or a snubber network (see the 9k schematic).

Nonetheless I'd prefer the suggested CM choke. I'd go with about 50 uH, add 100 ohm resistors in parallel (again to control peaking) and use two 1 nF capacitors to chassis afterwards. Farnell 926-5783 or 926-5791 look like suitable parts for the CM choke.

Click to expand...

Here's the 9k input stage:



The cut-off frequency of your suggested filter would be a lot lower than the 9k filter, or the filter in my schematic. Still far above the audio range, but why?

Regarding loading: you could try a discrete buffer inside the output stage loop if you don't want another opamp.

Click to expand...

I didn't quite get the "inside the output stage loop" part. What do you mean? I have considdered just using an emitter follower after the 1512 chip.

[quote author="JohnRoberts"]
I typically target -3dB change between pan pot centered and hard panned L or R. The values shown only change about 1.5 dB. There are probably multiple combinations of values that meet that criteria but 4x 13k resistors around a 10K (linear taper) pot works if my early morning calculations are accurate. I also show about 10 dB of loss so you need 10 dB of makeup gain in following stage.[/quote]
Yes, four times 13k looks fine to me also. I hadn't noticed your 43k note at first, but I got the same number. So either it's correct, or we're both wrong :grin:

PS to maintain proper polarity from input to output you can just swap +/- inputs for odd number of inversions. In a console you also want to be sure all inserts and such are in proper polarity wrt inputs.

Click to expand...

No inserts and stuff needed here - I have to keep it small and simple...

Thanks for the help everybody!

Best regards,

Mikkel C. Simonsen

I didn't quite get the "inside the output stage loop" part. What do you mean?

Click to expand...

I had the INA103 in my mind, with the THAT part it's not possible.

So far I have only used a pair of ferrite beads, so I don't think they do much harm (or have much effect for that matter). The peak would be at a very high frequency at least.

Click to expand...

Yeah, the peaking is in the MHz range, but that's exactly the problem as it will make your circuit more (instead of less) sensitive to RFI (at least at this frequency) and may cause the stability problems I mentioned. If it were at 200 kHz, it wouldn't matter. Just run a few simulations and you'll see (or ask me to post my results).

In addition to this, I found them to be picking up e.g. diode-switching noise (from the PSU). In a recent design, bypassing them made the noise drop 1.5 dB due to this effect.

The cut-off frequency of your suggested filter would be a lot lower than the 9k filter, or the filter in my schematic. Still far above the audio range, but why?

Click to expand...

Didn't get your question? Lower frequency offers better protection, right?

Samuel

[quote author="Samuel Groner"]

So far I have only used a pair of ferrite beads, so I don't think they do much harm (or have much effect for that matter). The peak would be at a very high frequency at least.

Click to expand...

Yeah, the peaking is in the MHz range, but that's exactly the problem as it will make your circuit more (instead of less) sensitive to RFI (at least at this frequency) and may cause the stability problems I mentioned. If it were at 200 kHz, it wouldn't matter. Just run a few simulations and you'll see (or ask me to post my results).[/quote]
I thought the frequency would be above the working range of the chip, but when looking at the number I guess it could actually be in the range.

The cut-off frequency of your suggested filter would be a lot lower than the 9k filter, or the filter in my schematic. Still far above the audio range, but why?

Click to expand...

Didn't get your question? Lower frequency offers better protection, right?

Click to expand...

Yes, but the closer you get to the audio band, the more likely the filter must be to affect the audio. When looking at different schematics most input filters have a very high cut-off frequency. There must be a reason (I guess) to set the cutt-off high. If not, everybody would just remove everything above 20kHz, or?

Best regards,

Mikkel C. Simonsen

Yes, but the closer you get to the audio band, the more likely the filter must be to affect the audio.

Click to expand...

Basically true, that's why I (and others) suggested that you replace the inductors with a CM choke. As the name implies, this part looks like an inductor for common-mode signals (i.e. interference) only, for differential signals (read: audio) it is virtually transparent.

When looking at different schematics most input filters have a very high cut-off frequency. There must be a reason (I guess) to set the cutt-off high. If not, everybody would just remove everything above 20 kHz, or?

Click to expand...

How did you derive the "very high cut-off frequency"? One of the main problems is that you don't have a fixed source resistance at hand and that makes the cut-off frequency (and Q of the filter) variable. And many "real" (not DIY) schematics I've seen do set the cut-off not all that high, e.g. Studer, AMEK 9098, Cadac etc.

Samuel

One non-DIY sample (where I have a schematic) would be the Rane MB1B. It uses beads and caps also. Here's the schematic:



But I agree the common-mode chokes are a good idea - they are also small, easy to get and not that expensive. And yes, figuring out the cut-off frequency is a bit difficult when you don't know the source impedance :grin:

Best regards,

Mikkel C. Simonsen

One non-DIY sample (where I have a schematic) would be the Rane MB1B. It uses beads and caps also.

Click to expand...

Perhaps they have beads with pretty low impedance? Duno...

And two threads I remembered on this topic:
www.groupdiy.com/index.php?topic=10471
www.groupdiy.com/index.php?topic=4317

Samuel

Now I finally have some time to work on this matter again...

Here's an updated schematic with some of the suggested changes.



Best regards,

Mikkel C. Simonsen

I didn't see it mentioned in this thread; I got a couple kits for this preamp from Mikkel, and they are very high quality with excellent documentation and affordable. Check it out:

http://electronics.dantimax.dk/Kits/Preamps_-_poweramps/index.html

Thanks a lot for the ad :wink:

Here's the complete (perhaps) schematic with phantom switch and peak meter. If I have made any major goofs, please tell me before I complete the PCB :grin:

I decided to put everything on one board, and simply "stack" 8-10 of them. I plan to just connect the boards with 10-way ribbon cables to keep it simple. All the unconnected pins will be grounded at one end of the cable.



Best regards,

Mikkel C. Simonsen