[ACMP investiupgradifications] All things PREAMP

[wmtunate]

hey alex, over at the homerecording thread, dgatwood said that the electrolytic caps on the V+ rail local to each filter board were different values in the 81 than they were in the 73.  I guess the 73 has them somewhere around 1000uF at each filter board, but the 81s are in the neighborhood of 100uF.

Also, he mentioned that the ground traces on the 81's filter boards were much smaller than those of the 73.  Where the 73 has a huge ground plane over the entire board, the 81 has tiny traces.

here's some of the relevent posts:
http://homerecording.com/bbs/showpost.php?p=3088534&postcount=7317
http://homerecording.com/bbs/showpost.php?p=3090116&postcount=7376

 

[Mylithra]

Today I was going to try your test of moving the Mid-hi board from the 73 I have that buzzes to the 73 I have that doesnt. For me, they both buzz with the EQ engaged and the Mid hi set to .36 but one does it far, far worse than the other. Instead of moving the Board, on a hunch, I moved the Transformer from one to the other. Orieted them the same as they were in the previous chassis and the buzz ended up moving with the power Transformer. On the one that use to buzz badly, (wearing a leather glove) I rated the Transformer around till it was almost all gone. On the one that didnt previously buzz but now does, I did the same thing, I could hear the buzz changing as I moved the transformer but It never really subsided, just changed harmonics. I do have some sheilding Im getting for the inductors and I anticipate that will mostly take care of the remaining buzz in the one that doesnt buzz, but I suspect Ill have to change the Transformer out on the one that does buzz now to get it to stop.

Something else I noticed. I had one on top of the other on my lap while testing these. I did notice the one that didnt buzz, started buzzing badly when I turned the other one on.

 

[alexc]

Yeah grounding noise and induced noise is a real Alice in Wonderland thing.

Move this and check that. Notice 'this' is good and 'that' isn't.
Fix 'that' Now 'that is good' and 'this isn't'
Repeat 1000 times until you become psychopathic.
Then 'learn to live with it'
Give up music recording forever cause you can't stand the hypertension.

Like the clap, best thing is not to have it in the first place.

 

[Mylithra]

Dont forget to wave the Voodoo Chicken Stick

 

[alexc]

>Dont forget to wave the Voodoo Chicken Stick

That's a given!

OK : another bit

- swapped the inductor from the good mid-hi board into the noisy mid-lo board and put it into the mid-lo position
  EQ IN, mid-lo all frequencies, max + gain : 50Hz harmonics -90dBu and some mid-hash  -80dBu
  tube shield on inductor reduces mid-noise hash by 3 to 5dBu

  The noise level is much lower now than with the mid-lo inductor

So, it appears :

  The mid-lo inductor itself is contributing to 50Hz harmonic noise quite a lot
  The mid-lo inductor itself is contributing a little to the mid-noise hash
  The mid-lo board    itself is contributing a little to the  mid-hash noise

I would guess that a 'better' mid-lo inductor would mostly reduce 50Hz harmonic noise
                              a 'better' mid-lo board would mostly reduce the mid-hash noise


Now to the other ACMP81 which is stock PSU and toroid but with mod transistors and reset wiring away from toroid

Line IP, set to unity gain IN and OUT (about -4dBFS on Cubase meters), bal I/O to Motu with basic noise floor of <-90dBu

- with EQ OUT, noise floor very low with  50Hz harmonics all well < -90dBu, no mid hash noise   

- with EQ IN, all bands 0 gain, OFF frequency  : very little change to 50Hz harmonics, no mid hash noise
- with EQ IN, Hi and Lo filters ON, all frequencies: very little change to 50Hz harmonics, no mid hash noise 
                    Hi        band ON, Max + gain, all frequencies  : very little change to 50Hz harmonics, no mid hash noise 
                    Mid-Hi  band ON, Max + gain, all frequencies  : very little change to 50Hz harmonics, mid hash noise <-90dBu 
                    Mid-Lo  band ON, Max + gain, all frequencies  : added 20dBu 50Hz harmonics, added 20dBu mid hash noise <-90dBu 
                    Lo        band ON, Max + gain, all frequencies  : very little change to 50Hz harmonics, mid hash noise <-90dBu 

- now sub in the inductor-replaced mid-lo board
  The 50Hz harmonics have decreased about 8dBu and the mid-hash has decreased about 5dBu.
  Has not decreased as much as before

So I conclude that :

- the onboard replacement toroid and PSU have contributed very little improvement except perhaps some on the mid-lo board maybe 5dBu or so.
- the wiring reset has also done very little
- the transistor mod fixes everything except the mid-lo band which stubbornly remains NFG
- the mid-band inductor is itself the main culprit on 50Hz harmonic noise and some mid-hash noise
- the mid-lo board itself is main the culprit on mid-hash noise and some 50Hz harmonic noise

Leastways, that's how *I* sees it.

Next - externalification of the toroid and PSU. This will probably take a few hours.

See you

 

[alexc]

And again ...

Externalification of toroid and PSU done. Here we go again. ACMP81 with transistor mods.
No change to ground scheme

Line IP, set to unity gain IN and OUT (about -4dBFS on Cubase meters), bal I/O to Motu with basic noise floor of <-90dBu

- with EQ OUT, noise floor very low with  50Hz harmonics all well < -90dBu, no mid hash noise   

- with EQ IN, all bands 0 gain, OFF frequency  : very little change to 50Hz harmonics, no mid hash noise 
- with EQ IN, Hi and Lo filters ON, all frequencies: very little change to 50Hz harmonics, no mid hash noise   
                    Hi        band ON, Max + gain, all frequencies  : very little change to 50Hz harmonics, no mid hash noise   
                    Mid-Hi  band ON, Max + gain, all frequencies  : very little change to 50Hz harmonics, mid hash noise <-90dBu   
                    Mid-Lo  band ON, Max + gain, all frequencies  : added 25dBu 50Hz harmonics, mid hash noise <-90dBu    ???
                    Lo        band ON, Max + gain, all frequencies  : very little change to 50Hz harmonics, mid hash noise <-90dBu   

FLYING F****G SON OF A B****

Subbed in the mid-lo board with the swapped inductor and the 50Hz harmonic noise decreases to about half.

Well at least it shows that the PSU and toroid aren't the problem regarding mid-lo 50Hz harmonics.
Externalising definately got rid of the mid-hash noise. I only have the 50Hz harmonics now.
Probably half due to the inductor, half due to mid-lo board.

Well - that's all folks.  - Your mileage will probably vary!

I'm now looking at mid-lo board layout tweeks.

 

[clintrubber]

I only have the 50Hz harmonics now.
Probably half due to the inductor, half due to mid-lo board.

Just an idea: as a temporary/troubleshooting utility perhaps an emulated coil (gyrator & caps) could be used ?

I didn't do the maths here, but this should be well feasible. It won't be the most time-efficient perhaps,
but could sure help for a structured approach to the troubleshooting (it'd skip questions like whether the coils are shielded enough or not)

Regards,

  Peter

 

[alexc]

Could be helpful - been a while since I did a gyrator.

I actually have a couple of Carnhill inductors - I may try one of these.

But I'm thinking that the issue is a layout  things now.
It doesn't seem like an induced noise issue as much as a good ol' on-board layout thing.

I'll probably start removing components in the hope of narrowing down the issue.

When I unplug the Gain pot, no noise. Unplugging it simply removes the whole filter
network from the feed-forward and feed-back path of the whole board, unless I read the circuit wrong.
This leaves the amp in circuit. So the amp seems OK. I'll do some signal tracing to confirm that.

It may wind up being a case of foregoing the use of mid-lo freqs 220Hz - 390Hz at anything more
than 1/3rd + gain. That reduces the effect to a sort of passable level.

On the upside, most of it works with the suggested transistor mod and I don't think external
power supply is really worth doing. I could live with the around -80dBu mid-hash this removes.

Please feel free to make any observations on my postings!

See you

 

[okgb]

How about publicly thanking you for all the work & effort

Thank you

Bummer as i often like to cut a narrow notch around 200 ish ,
but [ not totally serious ] perhaps a passive mains notch ,
could we get it steep enough , Ha !

 

[alexc]

No worries Greg - I just wish I could find a solution, tho'. 

So far the 'Zmix' solution alone has done all the good stuff.

I've just been following in the foosteps of dgatwood and others on HomeRecording regarding grounding problems,
as suggested earlier in this thread.

I can certainly confirm his observations about high-resistance ground wiring across the unit. I too see 6mV DC
and up to 80ohms to signal star along the ground wiring.

So now I'm doing as he suggests and upping the filter cap on each board as well as adding a heavier guage
ground wire from the ground of each board to star ground.

I've reduced the dc voltage and the resistance but not yet checked with analyser to see if it has helped any.

I'll have some more results tomorrow. Day 2 of work *finished*

Bye for now

 

[alexc]

Some more...

Upped the +24V filter cap from 100uF 35V to 1000uF 35V as well as added heavy guage grounding
wire from the -ve cap leg (ground) to star signal ground on each of the EQ boards.

No change to noise on mid-lo board. 

Noticed something odd.

The noise level at output of unit (contributed almost exclusively by the mid-lo board) reduces
as you increase the input gain selector to max. 

At max input gain, the mid-lo noise is almost gone.

Am I missing something here? Why should the noise level decrease as you turn up the system volume?

 

[okgb]

It's bad gain staging that's just attenuating the full signal ?
maybe there's a path that loops to ground or has more resistance at high att and raises [ or lowers ] as the gain switch is turned up

Talking out of my hat cause i don't have the schematic [ where i can find it ]

 

[MATTI]

I think 81 schems were here some months ago, or if not I can send them even if I can´t host them

Matti

 

[alexc]

The preamp Line Out noise (in this test almost all lo-mid noise contribution) reduced when the IP Gain is increased.
Also when the signal input is increased at the generator.

And another ...

I removed the inductor from the mid-lo board and jumpered the pcb inductor holes at pin 1 and 5.
This has the effect of allowing me to keep the filter in circuit but remove the the inductor.

Obviously it is not  filtering the same anymore, but it does keep the gain  pot, resistors and caps of the filter in circuit
(for Frequency positions 2,3,4 200Hz, 270Hz and 330Hz - the worst ones)

No noise. I checked that the gain knob still boosts as before (but not in the way of an RLC filter , just RC)

So that indicates amp is OK, the filter bank is basically OK and not introducing any inherent ground or induced noise.

To me, it's the inductor. It just likes to buZz or HuuMM or whatever, at least in the presence of this board layout,  in this unit.

I'm now subbing in one of my Carnhill VTB9047 inductors. It's not shielded or very special, but it is the one used in other home brew 1081s.

Later..

 

[MATTI]

You are getting there, it seems 

Matti

 

[alexc]

Finally ....

Carnhill inductor doesn't buzz (excessively).

That's right - mid-lo inductor board with Carnhill VTB9047

220Hz, full + gain, 50Hz harmonics <-90dBu and mid-hash <-90dBu.

Checked carefully lest mine eyes deceive me.  ;D

This is in stock ACMP81 with the EQ boards Q4, Q5 transistor mods only.
Same toroid, same PSU, onboard.

To be sure, it still adds some noise, overall about the same now as the mid-hi. Perhaps 10dBu above converters.
But definately not the same madness as before.

I'm now going to clean things up and put in permanently and do some more testing,
but I think it's in the ball park now, where things like external toroids and sheilding
and ground improvements will have some reasonable impact.

Disclaimer - these things have a way of changing inexplicably, so I'll monitor for a while and report back.

See you

 

[MATTI]

So when you upgrade these with Carnhill transformers you should think of the inductors also

Matti

 

[alexc]

Here is pics

More testing now.

 

[okgb]

I guess Colin is going to have a run on those !
nice work

 

[alexc]

Thanks Maxwall

I'm glad to add to the efforts of everyone on working out how to polish these puppies.

--------------------------------------------

My opinion on mods to the ACMP81 is :

1. replace the Q4 and Q5 transistors on all 4 EQ boards to BC441 and BC461
2. replace the Lo-Mid EQ board inductor with a Carnhill VTB9047 or equivalent

Based on my observations, this should  take the unit to a minimum working level and I expect <-80dBu total
noise with all bands engaged and max boost and individual spectral components <-85dBu.

--------------------------------------------

External mounting of toroid and PSU I expect would give another 5-10 dBu improvement.
I can't comment on whether the BC441-6 and BC461-6 variants will improve because they are not available anywhere I can find.

Further investigations I'm doing are :

1. check mic input
2. check 2N3055 bias
3. check for distortion
4. frequency sweeps at the EQ bands and gains
5. effect of using the correct values of mid-lo band inductor resistors  R18-21 (3K0,3K3,3K3,3K6) instead of ACMP 3K3 for all


Finally, I will "taste of sweetness sense of music" and  "let each note be injected into my soul"
   
So far it feels like this unit has injected something, but it ain't sweetness. And not into my soul!

I'll post my results as I get more.

Suffice to say, I feel a whole lot better.  But I'm still cautious.