Soundcraft Spirit Live (1990) mic input gain reduction mod

[mihaelbele]

Hi,
It seems that mic preamps have +10db gain when the gain pot is in it's zero position. Is there a mod to reduce gain to unity at pot zero position? Or even implement as trim/gain function? I think that the mod would affect the line input gain too, so I would probably leave few pairs of inputs as they were.
I'm mostly using electrets, and can only resort to resistive pads if mic capsule has a built in FET, but even the AKD D112 gets clipped when used in bass drum.
I've found a channel schematic this thread: link
Thanks!

 

[JohnRoberts]

Try reducing the values of R17 and R18. To drop the gain 10 dB you want to reduce the 12k to 1/3 or roughly 4k.

Maybe only modify one channel first to see how it works, before doing more.

JR

 

[ccaudle]

Try reducing the values of R17 and R18

Couldn't you also increase the value of R16? That looks like the classic instrumentation amp front end where decreasing resistance increases gain, and R16 sets the minimum resistance when the pot is all the way at one end. Changing R16 has the benefit that you don't have to worry about matching the replacement resistors (although for that class of device they may not be matched super closely to start with).

 

[JohnRoberts]

Couldn't you also increase the value of R16?

No.. R16 establishes the maximum gain not minimum gain (OP's complaint). You would need to replace the 5k gain pot with a larger value which also involves an unusual non-linear taper making them even harder to source.

That looks like the classic instrumentation amp front end where decreasing resistance increases gain,

It's commonly known as a "Cohen" topology.

and R16 sets the minimum resistance when the pot is all the way at one end.

minimum resistance commands maximum gain...

Changing R16 has the benefit that you don't have to worry about matching the replacement resistors (although for that class of device they may not be matched super closely to start with).

I would recommend using 1% resistors for R17 and R18, I didn't suggest a specific value because I don't remember them all. Should not be hard to find.

JR

 

[ccaudle]

No.. R16 establishes the maximum gain not minimum gain

Thanks, dangers of posting before the coffee brews.

 

[mihaelbele]

Thanks guys!!!!
I'll try it over the weekend and let you know

 

[mihaelbele]

Thanks guys!!!!
I'll try it over the weekend and let you know

It works! Tested with linkwitz modded WM55 into Zapnspark's Schoeps generic electret front:
12k || 5k6 (piggyback) = 3k8 results in ~6db reduction
12k || 5k6 || 5k6 = 2k3 results in ~9db reduction
12k || 1k8 = 1k6 results in ~12db reduction
12k || 1k8 || 1k8 = 840R results in ~14db reduction and 2db headroom on soundcard input (clipped at capsule or mic preamp)

WM61 handles higher SPL, but with more than 20db of input gain, mic preamp/capsule noise starts to be noticable on peak meters, so I'll try to match that with -20db reduction (more than that would probably degrade S/N ratio). If these resistors are bridged, will preamp operate normaly (or, how low can I go)?

PS. I got +10db number (in my first post) by comparing Spirit with my SSM2019 based mic pre at unity gain. If that comparison was correct, that would mean that when R17/18 are lower than 2k3, preamp actually trims the signal?

 

[mihaelbele]

I need a bit more help, please.

This mod introduced some sort of oscilation that looks like shifting resonant peak. When the desk is turned on, the resonance slowly moves up/down the whole audible frequency range (on modded channels only). As time passes, it seems that the shifting speed slows down, and that frequency span becomes more narrow, eventualy stabilising across very narrow band after a few hours. Peak amplitude is about the same on all the affected channels, but the peak frequencies are not the same/correlated. I used 1% resistors. This is with no microphone connected. When mic is connected, more resonant sweeps appear (some with wider q), and the sweeps are faster.

If it's too complicated to reduce preamp gain, would a 10k pot between R34 and gnd, with wiper connected to send/tip, work ok for variable reduction before the soundcard input?

Thanks

 

[JohnRoberts]

That old schematic link is not working now... The Cohen topology should be pretty stable. If you reduce the feedback resistors around the op amps too low that increases the open loop gain working with the Rs connected to the device collectors and op amp - input. Try reducing the value of those collector resistors a commensurate amount to the reduction in the op amp feedback resistors.

I have never seen this instability before, nominally in the Cohen topology the collector load resistance is similar to the feedback resistance for a unity feedback factor. Making the feedback resistors smaller than the collector load effectively introduces voltage gain into the feedback path.

Of course there may be other remedies but that is where I would look first.

JR

 

[k brown]

Why can't you just put an inline pad between the mics and the mixer?

https://www.shure.com/en-US/products/accessories/a15as?variant=A15AS
https://www.ebay.com/itm/2753692796...puhInn34r+d3l9wjGo7WdcNFM=|tkp:Bk9SR4zIwKmqZQ

 

[JohnRoberts]

Reducing the gain of the preamp should be lower noise than padding down the signal before boosting it up again, higher than desired, but with hot signals the preamp ein may not dominate the noise floor.

JR

 

[JohnRoberts]

Regarding the above schematic (that must not be reproduced or disclosed), the 12k feedback resistors and 5.1k collector resistors nominally attenuate the open loop gain more than 6 dB helping stability. BUT the 22 pF capacitors across the 12k feedback resistors are actually adding some HF boost to that open loop gain. This may actually be working against stability (for stability we want the open loop gain to drop below unity gain by the time that the phase shift caused by internal time delay reaches 180' making negative feedback positive).

See if removing the 22pF caps improves the stability.

JR

 

[mihaelbele]

Thank you all for your replies.

@k brown
As JR says - SNR, but I'm basing this on few assumptions which may be wrong:
- mics to be used on these channels are "noisy" (specifically wm55 & wm61) and can only use resistive pads between mic and mixer (which probably also brings down mic noise together with the signal)
- 20db resistive pad that has "correct" impedances may require too much gain compensation, and spirit becomes noisy on higher gains (I noticed that with ribbon microphones)
- 6db & 10db resistive pads do not have "correct" impedaces and may "color" the sound (actually, if they do, I havent noticed that with those capsules - or did not care at that moment)
So, i thught that the "simplest" and most versatile solution would be to shift the gain range on few channels from 15-60db to let's say ~0-40db.
Spirit's "clean" gain (noise does not raise above the soundcard's noise flor) is about 15db (probably total ~30db over unity), so resistive pads would probably be OK if not for impedance missmatch (I'd love to learn more about that in context of transformerless mic/preamp) and resistors' thermal noise (if this is even an issue in this case).

@JR
I will try that, and if doesn't solve the problem or it changes the sound in some way that doesn't sound right to me, I'll give up - it's just that it takes too much time to experiment or potentionaly introduce a problem that I'll become aware of in the middle of some other thing. But thank you very much for taking your time and looking into it, I appreciate it.

The thing that bothered me was clipping the soundcard input (connected via mixer insert point) - I should've started the topic with that piece of information. It seems that mixer has more headroom/different operational level than the soundcard (+4/-10 thing?) - the insert sends can clip the soundcard input while mic preamp is not clipping yet, so some attenuation will be required even after the mic preamp gain reduction. If mic pre can take hot signal and that same signal output clips the soundcard, then output needs to be padded, which also reduces the noise perceived by soundcard (if it's above soundcard's noise floor).

I've tested two things...
Increasing resistance between the emiters (extra 10k brings down gain by ~6db, and it seems that it does not change the sound). I can put this resistor on a switch on a few channels (eg. on/off/on for 6/0/12db).
I've put a 10k pot before insert send point (as voltage divider), and it seems to work ok without "coloring" the sound. What suprised me is that when ground is disconnected from the pot, leaving 10k in series, the volume drops by ~6db - is that series resistance making a voltage divider with soundcard's input impedance?
I still need to test how much padding is required between the mic pre with full scale mic signal and soundcard's input, and figure out if it would be reasonable to pad all the channels (eg. if spirit's "clean" gain can compensate for this padding without introducing noise).

 

[k brown]

FWIW, back when I was DIYing mics with Panasonic capsules, I powered them with a 'battery box' that had unbalanced outputs for use with portable cassette and later poratble DAT machines. I went to a July 4th fireworks show and found myself right next to the mortars. The sound was so loud that no setting of the rec level was unclipped, so I reconnected them to the recorder's line input and got a perfectly clean recording. Proof that the capsules themselves can take enormous SPLs.

If you have a way of powering the Pannys that doesn't require phantom power, try connecting them to the mixer's line inputs.

 

[meverylame]

The mention that 10db of gain sounds clipped with a D112 inside a bass drum is a bit of a red flag for me. I would very much encourage you to look everywhere else to see if you have gain or headroom issues. Not to sound remedial but have you confirmed the principal voltage of the consoles power supply at both the supply and locally at the module/input?

 

[mihaelbele]

@JohnRoberts
Removing 22p caps across R12k feedback resistors helped in reducing the number of sweeps to just one (narrow) when mic is connected. However, while doing SNR tests, it turned out that all channels have this type of sweep noise burried under hum and hiss, and that this mod brings it up to very much. Also, the whole mixer happily sings @150Hz (dominantly) and other 50Hz harmonics (without mic connected) When mic is connected, 50Hz becomes much more noticable, but I presume that's just poor capsule shielding on my side.

@k brown
I built those mics >10 years ago using Zapnspark's schoeps schematics

@meverylame
Thanks, headroom solved by padding the insert send / sound card input. Spirit power supply is showing correct voltages (+48V,+/-15V as shown on PSU label), but what's weird is that schematic shows requirement +/-17V... anyhow, probably the caps are bad, and probably the transformer needs to be potted (I'm blaming PSU for humming issue) - pluging the insert jack brings the noise up by 2db, turning on the power brings the noise up by another 4db (150Hz), connecting the mic brings up 50Hz hum (but that does not change overall noise level which is dominantly 150Hz and hiss).

Other tests confirmed that there's a difference in operating levels between the mixer and the soundcard, and padding the soundcard inputs by ~12db brings mic preamp clipping a bit below soundcard input clipping. 12k resistor in series with gain pot reduces gain by 12db, which solves the problem of transient clipping for wm55a which is one of the reasons why all this started (btw opening the connection between emiters reduces the gain by ~18db). Spirit has enough "clean" gain to compensate for all the padding. There were no noticable bad side effects by doing any of this, and I can now decide if I want to push transients into mic preamp clipping (either by using gain or mic with hot output) for ~9-12db before soundcard input clips.
All tests with inline resistive pad were compromised because it turned that one wire snapped.