[ACMP investiupgradifications] All things PREAMP

[0dbfs]

The maximum field on the transformer is directly below and above it and it measures about 100 milliGauss at the bottom of the transformer (opposite the lead break out).  This is pretty much in line with an off the shelf design, which is very cost sensitive due to competition.

Since the highest concentration of stray flux is on the top and bottom of this transformer, I was thinking that the bottom of the steel chassis will conduct magnetic lines of force.  I am wondering if perhaps part of the problem of the inductors and/or potential loop areas in the PCB picking up hum may be due to the steel box acting as a giant pole piece to conduct the stray flux from the transformer to other areas where it causes grief.

FWIW, when I was testing proximity of the toroid on my 81's I found that if I tilted the very front of the trafo up 1/4"-to-1/2" that the inductors picked up much less of the field. It was still pretty bad though.

It was as if the lobe of the field was somehow more focused on the LM inductor when the trafo was parallel to the chassis and focused elsewhere when slightly angled up to the front..

Cheers,
jonathan

 

[redddog]

Hey Steve. Are these fixes you're sussing out going to be available for the DIY crowd. These things are mother heavy and sending 8 of 'em to you from NY could be frickin' expensive. In another post, you'd mentioned doing the installs yourself and I was wondering if you were thinking that was the only way.

Thanks again for all your work.

 

[Steve Hogan]

Hey Steve. Are these fixes you're sussing out going to be available for the DIY crowd. These things are mother heavy and sending 8 of 'em to you from NY could be frickin' expensive. In another post, you'd mentioned doing the installs yourself and I was wondering if you were thinking that was the only way.

I believe I will be able to make the tranformers available as a separate item along with a handful of MUST install power supply parts. At present the procedure that I'm considering includes the following items:

1. Remove old transformer and voltage selector from box.
2. Remove heat shrink and unsolder old transformer from selector switch.  Install Heatshrink over primary leads, solder primary wires to switch entering from top and bottom and/or the side to provide clearance for larger transformer.  Remove Molex shells from old transformer and transplant them to the new transformer leads with molex pins already attached.  Attach green faraday shield lead of new transformer to a circuit ground point yet to be determined.
3. Remove PS PCB.  Replace 48V supply Main filter cap with 100V cap (63V is too low).
4. Probably replace 24 Volt main filter caps with 40 or 50 Volt units instead of 35V (too low again for High line conditions)

Optional, but highly recommended additional power supply improvements include:

5. Add 2 protection diodes to each LM317 24V regulator
6. Change 10uF ref cap to 100uF (fixes rising HF noise in 317 regulators)
7. Upsize the output caps in the 24 Volt supply.
8. change 48V regulator to output 52V (not sure best way to do that yet).
9. Replace 1N400x diodes with Diotec-USA SRP-206 Fast, soft-recovery rectifiers. This eliminates tiny bursts of RF that occur when the slower,  standard recovery rectifiers turn off.
10. Add RC Network across at least 1 transformer secondary to damp the power tranformer when the rectifiers shut off.  This further reduces any spurious emissions due to tranformer ringing.

Other mods include:

1.Replacing the problem transistors (or other fix yet to be determined to fix oscillations in the EQ stages)

2. Modify the gain switch to eliminate the positive feedback pop. I like moving the resistor instead of cutting the trace so far.

3  Replace the 1K series resistor in the phantom circuit with 200 Ohm (1k will starve high-current mics, especially with only 48 Volts from the supply. The 100uF/63 cap should be upsized to 220uF/63V and the phantom switch should be grounded when not in use to discharge the phantom feed cap thru the 200 Ohm resistor when phantom is turned off.

4. Check and/or correct polarity from XLR to XLR.  Some units have line inputs wired backwards.

5. Likely that 470uF instead of 100uF decoupling caps on the EQ stages may improve LF stability.

6. Other mods yet to be determined once I get these units in house.

All these things can be done by an experienced DIY, but for those that aren't handy, I was planning to do all that stuff for a fee.  Most likely I will offer a complete kit of parts with all the right caps and parts and the transformer in a bundle.  I haven't yet seen a box in person, and we are just now making the cores to make the new transformer prototypes.  Watch this space as I figure it out.

 

[recordinghacks]

Chance sent me the ACMP-73 schematics.

Top of page: http://recordinghacks.com/tnc-audio-preamps/

 

[okgb]

Looks like Tom at Cinemag will have a good price for us
on the inductors [ 20 ish or less ]  stay tuned [ in  ]

regards Greg

 

[Steve Hogan]

...we are just now making the cores to make the new transformer prototypes.  Watch this space as I figure it out.

The cores for the new transformers were wound this morning, are being annealed tonight and will be coated tomorrow.
I expect to wind the first articles tomorrow.

Today I opened up Matt McGlynn's '84 unit.  I am amazed at the general quality of the parts used in this preamp.
One would be hard pressed to make the chassis alone with the machined front panel for the money that these cost.  I see decent PC boards, a lot of decent parts including Wima and Roederstein caps. Most electrolytic caps are Rubycon 105C types.  My initial impression is that there is an awful lot of stuff in this box for the money.  All this is before I powered it up. -- Lighted pushbuttons -- impressive. there is no way you could even buy the parts in this box for what they cost.

OK, so my initial impression is that this box is well worth putting some additional money into it to really upgrade it into a very decent mic preamp.

Here's what I found so far in the power supply.  First of all, the Power supply PCB layout is OK on the 48V and 24 Volt outputs.  The Bipolar 12 Volts had some ripple which doesn't immediately make sense, since the PCB layout appears to be OK.  There is one connection that doesn't seem right to me.  The 24V and 48V supplies are grounded to the chassis via the screw nearest the power transformer. The Bipolar 12Volt ground, however, is attached to a different chassis standoff.  This may cause a problematic loop, because the Bipolar 12 Volt supply does connect to the 24 Volt circuitry and the grounds should be grounded in the same place.  It looks like the ground connectin is made with a zero Ohm resistor (jumper) so it looks like it may be easy to get the bipolar supply on the same ground as the rest of the preamp.  More experiments needed.

At 120Volts input, the Phantom supply filter cap has about 61 Volts on it.  At 10% high line it measures 69Volts and the cap is 63 volts.  I will be replacing that 100uF/63V input filter cap with a 220uF/100V Nichicon HE long life, low impedance 105C cap.  I will be using some Nichicon HE series caps and some Panasonic caps in this project because the Panasonic FM caps I usually use are not available in all the values/voltages/case sizes that I need to properly fit in the power supply PCB.  The Nichicons are excellent and inexpensive.  The main filter caps in any power supply take the most abuse due to the high-current pulses from the rectifiers and the heating caused by the ripple current, so these premium (but still very reasonably priced) caps are the right choice in this location.  The Factory Rubycons are 105C but are not the long-life types.  The remaining factory Rubycons don't get stressed nearly as much as the main input filter caps, so they will work fine when they are just acting as filtering on the regulated supply.

The Filter caps on the 24 Volt supply run at 34 Volts at 120 Vac line and reach 39 Volts with 132 Vac high line.  I will be replacing these 35 Volt caps with the long-life 105C 50 Volt parts, which is the next voltage up from 35.  There will be less total capacitance on the input filter caps, which is actually a good thing for reasons I will explain below.  On the 84 (and probably the '73) there are two 2200uF/35 Volt filter caps (4400uF total) which will be replaced with a maximum of three 1200uF/50V (3600uF total).  This is actually way more than is necessary due to the fact that these caps are pre-regulator.  Having really high capacitance in this position makes the current pulses from the rectifier extraordinarily high, which may be another factor in the hum being picked up by the circuitry.  Smaller filter caps have more pre-regulator ripple, but the current pulses are longer and lower in amplitude.  It is a common mistake for those who are trying to upgrade their gear that they increase these main filter caps to many times the original values.  This is really hard on the transformer and rectifiers because they must deliver the current in very, very short, very, very high current bursts.  Usually it's a lot better to use medium capacitance here to have reasonable  pre-regulator ripple, but much lower amplitude current pulses. 
In audio circuits, IMHO, the best place to have large capacitors is post-regulator.  The '84 supply has huge (4400uF) capacitance pre-regulator, but only 220uF/35V after the regulator, where huge would be good.  I will be upping that capacitor value to at least 470uF-- maybe even 1000uF if I can make it fit nicely.

Bottom line is that I am going to recommend changing out about half the caps on the Power supply board to run it more conservatively and to protect it from accidents.  My philosophy is that quality sound starts with a solid power supply, and this one can be made quite solid by simply changing out a few caps and adding a few extra parts.

I have a new question for 81 owners:
  On the 81 power supply,  the photos show a 4700uF/35V electrolytic on each of the two 24V power supplies.  these caps are lying down on the board.  Can someone please look to see if there are places for 3 total filter caps under the cap that is lying down, similar to the '84 and '73 power supply PC boards.  This cap must be replaced as well, because 35 Volts is not safe as an input filter cap with as much as 39 Volts on it.  I want to place an order for replacement caps tomorrow, so a prompt answer by someone on this forum will be greatly appreciated.

Stay tuned . . .

 

[okgb]

Hey Steve
if you can post pix of the progress that would be great ,
facinating to see what really makes things work
[ i;e good design  ]

 

[crazydoc]

I have a new question for 81 owners:
  On the 81 power supply,  the photos show a 4700uF/35V electrolytic on each of the two 24V power supplies.  these caps are lying down on the board.  Can someone please look to see if there are places for 3 total filter caps under the cap that is lying down, similar to the '84 and '73 power supply PC boards.  This cap must be replaced as well, because 35 Volts is not safe as an input filter cap with as much as 39 Volts on it.  I want to place an order for replacement caps tomorrow, so a prompt answer by someone on this forum will be greatly appreciated.

There are no other mounting holes other than the two for the large caps, C9 and C14. Board is blank underneath them.

 

[Steve Hogan]

I will try to shoot some pics, as requested, but right now it would consist of my hand moving my meter test leads and my oscilloscope probes around.

Testing the '84 unweighted noise level as delivered:  Tektronix AA5001 THD analyser, Level, RMS detector.
Preamp settings: (These settings were what gave me the highest hum level.  Other settings were better.)
Gain = max (-80)
Line/Mic = Line (no input termination -- just nothing plugged into either mic or line input).
EQ in , HI Q in
All pots to max, both EQ and Output Gain
All EQ's OFF except low shelf 220 and mid peak 0.36kHz
Scope trigger = AC line.

Clearly the hum is being induced by the power transformer. High current magnetic pulses are clearly visible in the output waveform.
As delivered:  Output level = -45.1 dBu  35 milliGauss magnetic field at mid inductor.
Power transformer turned upside down --  leads up instead of down = -51.5 dBu = 25 milliGauss
Power tranformer hung outside box as far as leads allow: -57.4 dBu = 7 milliGauss
Whether or not the EQ circuit is especially sensitive to stray magnetic flux, it is clear that lowering the stray flux will definitely lower the hum.  There may be other steps that can be taken to lower the sensitivity to stray field, and that will also lower the background hum, but lowering the field will directly reduce the hum.  By the way the field strength meter drops to zero when I turn off the unit so there is no other significant stray field on the bench.

I believe that the hum induced from the peak currents being drawn from the power transformer will be reduced when I reduce the value of the 24 Volt supply input filter caps.  I know that it seems counter-intuitive -- smaller cap = less hum, but because the very short, high amplitude magnetic pulses are inducing the hum, making it more "buzzy", the smaller filter caps will help lower that peak amplitude.

To verify my theory, I removed one of the 2200 uF filter caps from the 24 Volt Supply, thus reducing the total capacitance from 4400uF to 2200uF.  The scope showed a significant reduction of the highest 120 Hz rep rate peak in the hum waveform.  And the RMS hum levels went down approximately 1.5 dB at all three transformer locations.  The field strengths were the same, but that is an averaging meter.  This means that losing capacitance when I change the filters from 35 Volts to 50 Volts will actually benefit the situation.  As a matter of fact, I may further reduce the size of those caps to an optimum value.

As first delivered, the ripple voltage on the phantom supply (100uF) = 300 mV PP
The ripple on the 4400uF 24 Volt supply was 110 mV (which is unnecessarily low) 500 mV to 1 Volt would be fine.  On the other hand, the PP ripple on the (too small) 100uF Bipolar 12 Volt input filter caps is 3.6 Volts.  There is significant current drawn by the 12 Volt supply -- all the LEDs etc.  I will be changing the input filter caps on the Bipolar 12 Volt supply to 470uF/35 Volts.  That will reduce the ripple to about 750 millivolts, which is about right.  Hopefully, I will end up with a more balanced set of cap values for the entire power supply, correctly sizing them to balance ripple and excessive transformer/rectifier current pulses. and increasing the size of the post-regulator caps, which will produce sonic benefits.  By the way, I can fit a 1000uF/35V output coupling cap to the 24V output with no problem, so that cap will get a 4.5 x increase in size over stock.

More later ---

 

[Steve Hogan]

There are no other mounting holes other than the two for the large caps, C9 and C14. Board is blank underneath them.

Thanks, Crazydoc, for the remote look-see!

Based on my latest experiments detailed in the above post, a single 50 Volt cap with reduced capacitance from 4700uF should actually produce less hum.  I will spec out a replacement that is the same case size, and whatever capacitance you get in a 50 Volt part.  Maybe even less, but I'm not there yet.

Thanks for the prompt response -- your 81 has not yet arrived.

 

[okgb]

Looks like we'll be able to reuse the little inductor cup
[ as long as the replacement is small enough , or it just
sits over the side ]

Anyone have the real Neve 1073 , 1081 schematic ?

anyone have a bunk acmp 73 to volenteer the inductors ?
the chinese schematic i have doesn't seem to state inductor values
[ only " L" labels ]

 

[rodabod]

Looks like we'll be able to reuse the little inductor cup
[ as long as the replacement is small enough , or it just
sits over the side ]

Anyone have the real Neve 1073 , 1081 schematic ?

anyone have a bunk acmp 73 to volenteer the inductors ?
the chinese schematic i have doesn't seem to state inductor values
[ only " L" labels ]

Re-using the cup seems like a good idea. See the Kubarth website for 1081 details. The Chinese '81 schematic notes inductance values, and Alex C measured these I believe. The '84 and '73 use different inductors obviously.

 

[Steve Hogan]

Looks like we'll be able to reuse the little inductor cup
[ as long as the replacement is small enough , or it just
sits over the side ]

Anyone have the real Neve 1073 , 1081 schematic ?

anyone have a bunk acmp 73 to volunteer the inductors ?
the chinese schematic i have doesn't seem to state inductor values
[ only " L" labels ]

Download the Neve schematics at the AMS/Neve website.  The manual for the 1073, 1084 and 1081 have complete schematics in the back.

Why, exactly are we re-doing the inductors?  Do we know that they are the wrong values/resistances, etc. compared to the original Neve EQs?  These inductors are pot cores, and they are inherently self-shielding (up to a point). If additional shielding is required once the power tranformer stray flux is reduced, I know Jensen sells empty annealed 80% nickel transformer cans that could possibly be slipped over the inductors right down to the circuit board, and held in place with a dollop of Dow 748 electronic grade RTV.  The 80% nickel material won't work in very strong magnetic fields because it will saturate, and once that happens, it no longer acts like a shield.

If accurate measurements need to be made of either original Neve or Chinese inductors, I have a GenRad 1659 RLC digibridge that can do a decent job.

 

[alexc]

Hi Steve,

>Why, exactly are we re-doing the inductors?

If you wish, you can have a look at the description I posted earlier in this
thread relating to basic noise identification tasks that I performed showing the noise
contributor being the mid-lo inductor.

Although it is hardly an exact analysis, I found it useful in focussing my efforts for
mitigating the performance issues I experienced with the units in hand.

I look forward to your noise measurements and observations on the 81 when you have
had an opportunity to examine one.

 

[okgb]

Alan Hyatt had brought up claims of this being a prior design
which he had something to do with and rejected ,
forwarning of inductor problems they had which seems to have
more credibility now , you'd have to dig through the threads
but if the chinese skimp on xfmrs what confidence does one have
in their inductors ?

And i don't know about anyone else but when i tried mine through
the hum , the actual Q button " felt " funny , maybe unrelated
and/or my imagination .

And thanks for the redesign work , who would think to reduce the capacitence !?

 

[recordinghacks]

Here is Alan Hyatt's comment about inductors:

When I read another post where it was said this project was a year long project, again I came in to make a correction.
Why, because I have seen these units, and have been involved in fixing resonance peaks on the inductors when we were going
to do it under the SP ENVY IN June of 2007, but I eventually passed on them.

In context: http://homerecording.com/bbs/showpost.php?p=2943222&postcount=4910

 

[jwk1]

I wonder if anyone could help me with a question about the values of the resistors across the secondaries of input transformers on the acmp73, when upgrading the transformers to carnhills?

The acmp73 has 2k2 (R2) on the line input secondary and there is 12k (R1) on the mic input.  From looking at the original schematic at

http://www.danalexanderaudio.com/neveinfo/Neve1073schem.jpg

the 2k2 on the line input seems to correspond to R1 on the original neve schemo.  However I can't find a resistor on the original schematic which corresponds to the 12k (R2) on the acmp73 schematic.

Should I conclude from this that the 2k2 on the line input should be kept, but that the 12k on the mic input should be dispensed with, when I swap transformers over?

Thank you for any help, as I am not too knowledgeable in these matters...

Josh

EDIT: I think that I just found the answer to this question here...
http://www.auroraaudio.net/dcforum/DCForumID1/131.html

 

[maxwall]

I agree 100% with the following



    Steve Hogan
1. The power supply must be made solid, this is the foundation of any recording tool.
    thanks for the re-assurance, testing  and improvements. I believe this
    issue contains about 80 percent of the noise issues and other design problems.


    Greg Boboski
2. The inductors must be investigated tested and fixed as needed.
    I suspect that if the inductors have the same level of quality and reasearch as do the
    toroid power transformers then this leaves me with absolutley no confidence in trusting they
    will perform as needed. However, this should be done after the power supply has been
    properly fixed.

 

[Steve Hogan]

Results of supplied ACMP Preamp Chinese Power transformer tear-down:

Actual core size is 1.1" ID x 2" OD x 0.6" High.  At 115/230 Volts (50 Hz) the flux density is 16.34 kGauss. At 120/240 Volts, which is much more likely in the USA today due to utilities maintaining higher voltages to reduce transmission line power loss, the 50 Hz flux density is 17 kGauss. As I predicted, this level is right at the maximum operating level of the core material.

The quality of the winding, taping, leading, was OK.  Taking off layer after layer of secondary windings was tedious and uneventful.

The real interesting part was the primary and the core.  It is clear to me, based on the construction technique used for the core, that the transformer manufacturer makes one standard 115/230 Volt primary in a certain VA rating (I suspect this is their 10VA rated primary).  I am very sure that they make these universal primary windings by the 10's of thousands all the same.  They then make semi-custom transformers by adding various secondary windings to the stock primary+cores to make whatever model transformer is required.  The primaries are wound for maximum utility of the core, minimum number of turns, minimum core size to accomplish the VA rating required when operating the core at maximum operating flux density.  Reasonable commercial practice, but not suitable for use around audio circuitry if you want low hum levels.

Here is the really interesting part.  Remember that in one of my earlier posts, I said that we had torn down some poorly-performing Chinese toroidal power transformers and found that they had made the cores out of multiple scraps of strip steel instead of one continuous strip of grain oriented silicon steel?  Well, because the primary of my sample ACMP transformer was varnished to its core, I could not unwind it, but I had to just cut and peel it away.  It was a waste of time to remove it all, so I removed a portion on each side of the core 180 degrees apart so I could get calipers on the core to measure its size.

In the one small portion of the core where I stripped it down to the steel, I found the end of one strip. about 0.125" in from the OD.  This core in this particular transformer was therefore made from at least two pieces of steel. If I tore the rest of the primary off, maybe I would have found other end gaps that would indicate more than two strips. What's the chance of me finding that out from uncovering 1/2 inch of one core from one transformer, if it were not common practice to make the cores from whatever scraps of steel they could find?  See the attached photos of the uncovered core and a close-up showing where the gap is in the core from where an inner strip of steel ends.

If the core quality in these Chinese transformers is a total crap shoot (pun intended), that may explain why some folks are having terrible hum from their preamps and others not so bad.  Even if you luck out and get a core made from a single strip of steel, the flux density is very high for including in a 1 RU audio preamp with unshielded inductors.

The cores in the transformer I will be making available is made from one continuous strip of grain-oriented silicon steel.  We wind, anneal, and coat the cores in house, so we maintain absolute control over the material used and the processing.  The photo below shows the Chinese core, the (green) core used to make the replacement transformer and the first prototype transformer that I finished yesterday, but have not yet tested in a preamp. The extra Green/Yellow wire connects to the copper foil faraday shield inside the replacement tranformer that keeps AC line noise on the primary from capacitively coupling into the secondary.

I also obtained, with the kind assistance of Dave Hill at Jensen Transformers, a collection of various sized hydrogen annealed, 80% nickel cans and covers, with which to experimentally shield the inductors in order to quash any remaining hum and buzz.

To be continued:

 

[mitsos]

Looking forward to seeing the results of this in a preamp. Waiting and hoping your toroid will fix most issues...

OT about the faraday shield.. just wondering about the process. Do you wind the primary, then wrap the shield, then wind the secondary on top?


Thanks again for doing all this!