Beag PKC-901 master bus noise improvements?

[12AX7]

Hey everyone!

I've got a Beag PKC-901-12 12 channel console (3 actually, 2x 16 and 1x 12) that I recently restored. It's a cool mixer and really well built. Has some pretty nice preamps. But it does have one pretty big issue - master bus noise.
All channels have direct post-fader outputs but I'd still like the master bus to be useful. Right now, with all faders down but master faders up about half way, the hiss is pretty audible. Push the faders up a few more centimeters and it gets louder. Push them up almost all the way and it's pretty unusable. And this is on all three consoles.
The board has been fully re-capped with high quality caps. I've also done some PSU improvements - larger primary filter caps, slightly higher filter capacitance after the regulators and 100n rail decoupling caps for each channel on the large bus PCB on the bottom of the console. This did not improve the issue a whole lot.

I've drawn out the schematic from some pic of the service manual, it's attached to this post.
Up top is what the channel output looks like from after the EQ section and into the master bus. The bottom is what each master bus summing part looks like. If master faders are down, there is very little noise.

What could be done to improve the situation?
Some thoughts I have:
1) Output resistors of each channel and input load resistors on master bus have to be re-calculated?
2) Gain of first summing op-amp stage has to be re-calculated?
3) Summing amp and channel outputs use 082D/084D op-amps. Replace them with 072B/074B?
4) All of the above?

Both the 12 channel and 16 channel versions suffer from this. There's also an 8 channel version but I've not found that yet so I don't know how bad it is. I'd imagine it's a bit better with less channels.

Any help greatly appreciated!

 

[JohnRoberts]

1- TL08x is actually noisier and worse spec than TL07x. Even TL07x is not good choice for post fader gain stage, or sum bus amp.

2- Besides using a relatively noisy bifet op amp for post fader gain stage, the values shown indicate > typical +10dB gain there. Gain of 5.5x instead of 3x will contribute more noise. I am unclear about spare parts feeding the - input pf that post fader gain stage, perhaps a trim to improve fader kill. It looks like the layout will only accept quad opamps limiting obvious options, using a TL074 and increasing R58 to 5k or so will get post fader gain to nominal +10dB.

For good news that pan approach is lower noise than typical lossy single pan pot.

3- bus amp is a dual so consider a NE5532 or modern lower noise bifet dual than TL072. 

4- the active LPF after the bus amp looks unnecessary.

5- the post master fader gain stage again looks like too much gain, make R26 more like 5k (5.1k, 4.99k), the trim going into the + input of U8A looks wrong... Again TL08x or even TL07x are not obvious choices to drive the outside world. Those bifets are only rated to drive 2k minimum loads.  That trim circuit looks like a 133 ohm load that bifet would never drive, probably a incorrect value or more.  Since that looks like another quad package not obvious what modern substitutes are available. Others may be more aware of modern choices.

======

Long story short pulling the excessive gain out of channel and master fader gain stages will help. Using lower input noise op amps in bus sum amp, will help. Noise of that one op amp is being boosted N+1 for number of channels feeding sum bus.

I see a build out resistor in the direct out (good) there should probably be some resistance in series with the master output (to decouple from external capacitance).

I don't have an obvious suggestion for good modern quad op amp to replace TL074.

JR

[update- Digikey only shows one audio quad opamp in through hole package in stock (LME49740), but it is indicated as obsolete.  Note: the supply current suggests that wholesale substitution might overload the stock console power supply, but maybe OK for a few critical sockets.  /update

 

[12AX7]

Thank you very much, JohnRoberts!

How about the TLE2074 as a replacement for the 084s? Or OPA4134?
For the 082s I was thinking 072Bs. You think the NE5532 would be a better? I have a bunch of those. Also some OPA2134s.

I re-checked the + and - inputs on the opamps on the original schematic and that’s how it is originally. Will have to look at the board if it really is like that. Same goes for that 133 ohm load, looks weird but it’s in the original schematic. Opamps in the bus amp are all dual.

Btw that master section has more stuff before the final output. Where I labeled output to etc, it goes into a fee more stages. I can draw those out as well.

I do agree with the excessive gaining. This board has some pretty hot signals running in it. I had a pair of M Audio BX5 D2s that I use for testing hooked up, volume half way up, running console preamps into max clean gain, channel faders at unity, I could push the speakers into clipping with master faders half way up!

 

[JohnRoberts]

Thank you very much, JohnRoberts!

How about the TLE2074 as a replacement for the 084s? Or OPA4134?
For the 082s I was thinking 072Bs. You think the NE5532 would be a better? I have a bunch of those. Also some OPA2134s.

I am reluctant to give advice about parts I haven't used myself... Over the years I used truckloads of ne5532

I re-checked the + and - inputs on the opamps on the original schematic and that’s how it is originally. Will have to look at the board if it really is like that. Same goes for that 133 ohm load, looks weird but it’s in the original schematic. Opamps in the bus amp are all dual.

If they are in sockets you could drop in a 5532 to see if it reduced noise.

Btw that master section has more stuff before the final output. Where I labeled output to etc, it goes into a fee more stages. I can draw those out as well.

I do agree with the excessive gaining. This board has some pretty hot signals running in it. I had a pair of M Audio BX5 D2s that I use for testing hooked up, volume half way up, running console preamps into max clean gain, channel faders at unity, I could push the speakers into clipping with master faders half way up!

If you don't need all that gain its not that difficult to reduce it some, 10dB is typical.

JR

 

[12AX7]

Thank you so much!

I'll try to figure out a quad opamp to get for the channel outputs.

For the duals in the summing amp I have some NE5532s and OPA2134s to pop in there. I should probably take some measurements before and after each set of changes to see what works best. All opamps in this board are soldered but I'm going to install the new ones in sockets anyways.

I drew out the rest of the master section as well. Hopefully all the values are correct. Do you think any improvements could be made here as well? Probably a good idea to replace all the TL08xx series opamps in the master signal chain? There's another set of 082s driving a 100 ohm load for the output drivers. If any values look really out of place or any connections look weird then let me know. I drew this off of some pictures taken with a smartphone of the original service manual. Unfortunately I don't have one and there is none on the internet so I have to rely on the pics I was sent by a guy who has the manual.

Pretty interesting devices driving the master outputs - Signetics NE540L

EDIT:
Had a bit of spare time today and curiosity got the best of me. So I pulled the mixer from my shelf, fired it up and listened to the noise level. I then pulled the master section and went ahead and changed the first two master summing opamps with NE5532s and R26 and it's other bus equivalent with 5K1s.

Fired it up and WOW! Noise levels are instantly improved! Without even any measurements I can say that this thing is quite a bit quieter already. All faders on this thing are sectioned 0-12. Master faders at 12 now have the same noise level as they had before on about 9-10.

I'll order some some NE5532s and DIP8 sockets and re-opamp the rest of the master module I think.
I've got the DIP14 sockets and 5K1 metal film resistors for the input channels. Now to figure out a quad opamp for them. I suppose I should also keep in mind the power consumption, right? If the few opamps in the master section consume a bit more, that should be okay. But higher consumption across all 12 channels might be an issue.
PSU is powered by 7815 and 7915s. Do they make higher current versions of them?

 

[JohnRoberts]

Glad it helped...

JR

 

[Bo Deadly]

That is a strange circuit. With all of the coupling caps, it looks like it was designed for amps that had some offset. But TL08X do not have enough of an offset to justify all of the coupling electrolytics. And TL084 is just the wrong part. I have to guess there was something else there and someone decided to swap in the TL084. Unfortunately I'm not aware of any really low noise quad PDIP14 amps. Also, do the amps have bypass caps? If not, you really can't use high speed amps like OPA2134. One option might be to make a "surfboard" that adapts two dual SOIC to the right PDIP 14 socket. Then you can add a bypass cap across the supply pins which isn't ideal but might be just enough to keep the amps stable. If you do have bypass caps, you can buy off-the-shelf surfboards on Ebay to use an SIOC amp like maybe OPA1664 (not familiar with that amp but searching Mouser it's a low noise quad that "looks the part"). And with a little minor soldering hackery you could add the bypass cap across the rails to the off-the-shelf surfboard.

Otherwise, I don't see any obvious reason as to why this circuit would be that noisy. Define noisy. Is it really pure hiss or is there any hum or whistle? The gain after the fader is not that much. It could be that they run the upstream circuitry running at a lower signal level to account for EQ headroom or because they found it just performed better that way. If you use a rail-to-rail amp like OPA1664, you might find that you can just take out R26 and make that stage unity.

It could also be that the channels are actually noisier than you think but you only hear the hiss when they get summed. I would look at the preamps. Any improvement there would be multiplied by the gain so getting the preamps to be as low noise as possible can make a big difference in overall noise performance. Post a schem of the preamp.

That weird "trim" after the fader is almost definitely a mistake in your schem. That load cannot be that low. As JR pointed out, the amp would have trouble driving it and you would get significant distortion. Especially with a lowly TL084. And it just doesn't make any sense. My guess would be that that net that you think is ground, isn't really connected to ground and it's doing something else. If it's not wrong, it needs to be removed ASAP. This circuit does seem a little over-engineered.

 

[abbey road d enfer]

Otherwise, I don't see any obvious reason as to why this circuit would be that noisy. Define noisy.

+1. Even with the pedestrian TL08x, the noise level should workable.
What is the set-up that leads to the conclusion that there is too much noise?
Measurements would be welcome.
Indeed, replacing the TL082 with 5532 or OPA2134 is bound to improve performance.
And making sure there are local decoupling ceramic caps is mandatory.

 

[12AX7]

Thank you all for your input!

On the bottom of this post is a Google Drive link with the console documentation. I've separated it into folders for easier navigation.

To clear some thing up - about that 133 ohm load, it was my mistake. The schematic photo is very hazy there, it was R46 there not R45. So it's not a 33 ohm resistor but a 56k resistor. Sorry for that! I've fixed it.

All the 08xx opamps are original. They were installed from the factory and are also specified in the parts list. My guess as to why they used them everywhere is that Beag, being a Hungarian company, in the 80s did not have access to a massive selection of foreign parts to choose from. They used some really great parts, though, like ALPS faders and gold plated Neutrik XLRs in this mixer. But sometimes they still had to go with whatever was available. This console is a mix of ICs from ST, TI, Motorola, Signetics and RFT (Germany). The TL084s are made by RFT and TL082s are made by Texas Instruments in this console.

About the noise - it's just pure hiss. Used to be hum as well but that was fixed after a full recap. This hiss is present in any setup. Whether the output is going into a pair of speakers or amp (tried with 3 different sets), an interface (tried with SSL2+) or running into another mixer (tried with Midas Verona 320, Soundtracs Topaz Mini, A&H Mixwizard 3), the hiss is very much present at any reasonable output level. Right now with the master bus mod, I had it running into my A&H MWZ3 in my workshop. All faders down on the Beag except the master out faders. On the A&H, all channels down except 1 and 2 where the Beag was plugged into. Line pad engaged, gain at minimum, channel faders at unity, output at unity, running into a pair of Tesla ARS 651 active monitors, also at unity gain. Master faders half way up on the Beag resulted in VERY audible hiss in the speakers. After changing summing opamps with NE5532s and changing R26 to 5K1, with the same testing setup, the noise was reduced a lot.

As for the console preamp noise - it's not the quietest board anyways BUT the noise on the master bus is there even with all faders down and all gains at minimum. So it's definitely not a preamp issue.
Mic pres are based around a TDA2310 and line ins are based on TL084 once again. Those should be replaced as well. The TDA2310s are actually quite interesting. They have a certain sound to them. Most people who have used these consoles have all said the same thing - the pres are actually nice and the whole board sounds good but everyone has retired them mainly due to the high noise. These were used in many studios here in Estonia for recording and broadcast in the 80s and 90s.

I don't remember if there were decoupling caps straight on the opamps but I can add them. PSU is set up so that there's a PSU module that has the transformer, rectifiers, main filter caps, phantom power generator. Then that unregulated DC runs on the bottom bus PCB to a small board with a pair of 7815/7915 regulators mounted in the middle of the console, on the bottom, onto a metal bar. That then returns to the bus PCB and each module has it's own local regulation where it's regulated to +-13.5V.

Quad op-amp choice is somewhat limited by that PSU setup. Using opamps with high current draw may be too much for those 7 series regulators.

Here's a link with the schematics:
https://drive.google.com/drive/folders/1NbI8lMoDDMqrFEqV68xAH7iR8I3p5gp8?usp=sharing

 

[Bo Deadly]

Yes, if it's hissy with the channel faders down, then it's not the channels.

I have a new suspect. My money says the outputs are a total disaster. R67 and R68 attenuate going into the NE540L which apparently has some significant fixed gain. It's hard to see exactly what the sense circuit is doing there but it's definitely a problem.

To prove it, rig a scope probe to a USB audio interface and find some audio analyzer software. This is the most basic rig that you'll need to do any sort of really quantifiable analysis. Then use that to look at the output, look at the spectrum and note how far down the noise floor is. You don't have to calibrate the level at this point to do comparative measurements. If you don't have a scope probe, just use a piece of shielded single conductor or thin coax with a very low ohm resistor on the end as a "probe tip".

Note: Because the probe is an unbalanced connection, you will want to connect the probe tip to pin 2 and ground lead to pin 3.

Then, and this is where we prove that it's the output that's noisy, attach the probe to the upstream side of the insert switch K2. Now look at the spectrum again. I think you'll find that the noise drops by some absurd amount. Yes, U10 adds some gain but adjust it to 0dB or use a tone to measure the exact difference. You'll still find that the noise drops by a huge amount.

In the end, if you really want this thing to be low(er) noise, you will need to seek and destroy every instance of NE540L and the circuitry it's connected to and fashion a new replacement daughter PCB that you can retrofit into there. I would just do simple impedance balanced outputs but if you want to get fancy you can use THAT 1646 outs.

It seems like this piece suffers from over-engineered circuits and IC supply problems. But it looks pretty cool and has a retro vibe to it. So if making a little PCB to replace the output makes it work, that could be worth the trouble.

 

[JohnRoberts]

Come on man....

TL07x are ballpark 3uV ein, TL08x are not even that quiet. The typical virtual earth sum bus has a noise gain of N+1 the number of stems being summed. So  17x 3+uV is more noise than 17x the 5532 ein (1uV or less ?).

Of course 16 noisy channels, mics self noise, room noise etc, could easily swamp that bus noise floor out.

Noise floor is just that, the absolute best case scenario and not real world operational expectations. 

Better is always better but I do not advocate throwing too much time and money at such pursuits. This was relatively low hanging fruit (easy/inexpensive). The next few dB will be harder and more expensive.

I spent a lot of time analyzing the calculus of actual console noise performance. The excessive gain in the channel post fader gain stages will contribute noise but incoherently so instead of the 17x for bus amp ein, more like square root of 16 (only 12 dB more than 1 channel).

JR

 

[Bo Deadly]

Well you would know and I agree that no matter how much work you put into it, it's never going to compete. A Behringer XENYX or Mackie VLZ can be had for like $200 on Ebay? But this Beag looks like the construction is actually sturdy and it's got a little vintage mojo. Some of the old Studers are really not that great and people drool over those things. If someone has the time, it shouldn't cost much to change things up and make it work well enough for someone to enjoy it for a personal or monitor mix. Small format vintage mixers are all the rage ya know.

 

[JohnRoberts]

Well you would know and I agree that no matter how much work you put into it, it's never going to compete. A Behringer XENYX or Mackie VLZ can be had for like $200 on Ebay? But this Beag looks like the construction is actually sturdy and it's got a little vintage mojo. Some of the old Studers are really not that great and people drool over those things. If someone has the time, it shouldn't cost much to change things up and make it work well enough for someone to enjoy it for a personal or monitor mix. Small format vintage mixers are all the rage ya know.

The pan topology is actually lower noise than typical mixers that involve lossy single pan pots, requiring post pan make up gain. While that is just one of several terms contributing to overall console noise floor performance. 

I wouldn't junk it just yet, or gold plate it.

JR

 

[gridcurrent]

in agreement with squarewave that the outputs are a disaster.
the 540 is famous for being problematic, oscillation is the norm.
the Dalcon console previously at Oceanway Studio B was designed wit 540's and despite its legendary sonics, the summing had HF noise.
attached is my suggestion for the output stage.

 

[Bo Deadly]

attached is my suggestion for the output stage.

Agreed. That's a nice solution. Impedance balanced and there's a good chance that it could be modified without adding a PCB. The feedback cap could be tacked on to the socket from the underside.

 

[12AX7]

Once again thank you all for your input! This is quickly becoming my favorite forum on the internet, so much to learn from here!

For the quad opamps, OPA1664 seems like the best option so far. I should have enough SOIC-14 to DIP-14 adapter boards and pin headers to make enough adapters. Should the decoupling on these opamps be rail-to-rail or rail-to-ground? Film or ceramic and what value? I have 10n, 47n and 100n ceramics in stock and also 22n and 100n film.

I do have a scope and measurement tools to measure the noise floor. But I didn't. But I should have before I began. Oh well...

Anyways, I'm going to explain the situation/goal with this console, something I should have done in the first post. Maybe this will help on deciding what's worth doing and what's not.
The goal is not to make it the quietest console in the world. It's also probably not worth chucking $10/a piece opamps into it.
It's not that I want to make this super quiet and squeeze every last dB of noise out of it, no. It was just to get the noise level down to a more usable range. The master bus opamp replacement and gain reduction already was a nice improvement. And I'll have to agree that the gain was a bit excessive purely on how hot the output was compared to any other console I have when ran with the same levels.
I use this mixer for preamps when tracking/mixing drums. Why? Because the preamps, believe it or not, really do sound cool. This console naturally has a massive low end to it for some reason. It doesn't sound boosted, it just sounds like there is more low end naturally and it reaches a bit deeper. And quite a bit of color and mojo. Overall, I love it. And I love the aesthetics and the build quality and feel of faders, pots, switches. Overall just a really cool and well built console.

The goal is to just get the self noise hiss down to a workable level. That's the only issue with it. All the electrical problems I've already sorted out. Some channels were not working, mostly dead opamps but I have three of these consoles - this 12ch, another 16ch waiting to be restored and one 16ch for parts. So I was easily able to pull anything I needed.

To sum up - is this the best console ever? Of course not. I own a few mixers, nothing high end but definitely better ones. Yet, this one is my favorite still. There's just something about it that appeals to me. And I don't need it to have a world class sound, I like the sound it has. It's just the hiss that's too much.

 

[12AX7]

in agreement with squarewave that the outputs are a disaster.
the 540 is famous for being problematic, oscillation is the norm.
the Dalcon console previously at Oceanway Studio B was designed wit 540's and despite its legendary sonics, the summing had HF noise.
attached is my suggestion for the output stage.

Thank you very much!

In this case the output would be unbalanced, right? Is there any replacement for the NE540L that would retain a balanced output?

Or maybe build a new board with the suggested THAT 1646. If I strip the NE540Ls out, the TL082s driving them and all the related resistors and caps, that should leave quite a bit of free room on the PCB. An SMD THAT 1646 should fit just fine onto a small board there.

 

[gridcurrent]

In this case the output would be unbalanced, right?

the suggested circuit is impedance balanced.

Is there any replacement for the NE540L that would retain a balanced output? 

yes.
fed from your newly installed 5532 (U10a), a 1:1 output transformer.

 

[12AX7]

I think I have some Lundahl 1:1s but I'm not sure if they would fit in there. I'll have to check though.

So now I have four options for the output stage:
1) Keep the original as it is and see how well it performs after the rest of the summing bus modifications
2) Use the impedance balanced circuit (very easy mod and a pretty cool idea)
3) Use a 1:1 transformer (easy mod but would have to find two that fit on the module)
4) Use a THAT 1646 based output stage (not the easiest but should be doable)

 

[abbey road d enfer]

Should the decoupling on these opamps be rail-to-rail or rail-to-ground? Film or ceramic and what value?

It has been demonstrated time and again that rail to ground is a better proposition.

I have 10n, 47n and 100n ceramics in stock and also 22n and 100n film.

use the highest value available (100n is generally good) and ban film caps for decoupling.