[ACMP investiupgradifications] All things PREAMP

[Steve Hogan]

Steve Hogan Preamp Modifications update:
User input solicited

I have spent a couple of days working to finalize the mods for the preamps.  I have taken more photos and have spent more time on the instructions.

As I have previously explained, I have yet to finalize a practical fix for the Gain Switch "pop" problem in the preamp section.

The elegant switch modification solution that Crazydoc did for his units is mechanically challenging and probably not for the vast majority of owners, so my proposed solution is to reduce the 11 steps to 10 steps and include an "OFF" position somewhere in the middle. The gain steps would then be spread out in order to cover the 60dB gain range in fewer switch positions. At the same time, it would be nice to optimize each remaining gain position for best performance.

The gain structure approach of these Mic Preamps using fixed amplifier gain with post-amplifier attenuation cannot optimize signal-to-noise ratio nearly as well as a completely different approach that would use variable gain amplifiers. Therefore this preamp is best used for microphone sources that are medium to high level. A practical modification must use the gain switch "as is" and change or move some resistors and/or cut a trace or two. The original Neve circuit used a gain switch with many more steps and also more poles so the gain of the amplifier stages are changed in addition to pads being inserted.
It's just a more complicated approach than what we have to work with here.  (By the way, even with its increased complexity, the signal to noise ratio is still not as good as other approaches).

The following techno-stuff is offered to solicite feedback on my two proposed solutions to the gain switch "Pop".

One approach has finer gain steps at high-gain positions and more spread out low-gain steps.
The other has finer steps at the low-gain position and is more spread out at high gain.

It is not possible to make all the steps evenly spaced unless one wants to give up the -50dB (or -51dB) step that has no attenuator. This is the first low-gain position which bypasses the first stage entirely.

Simply eliminating a gain range in the middle makes a very large 12dB jump in the middle.
Fixing this in a way that works really well is not easy.



The original ACMP mic preamp circuitry used a first stage gain of 30 dB and 5 steps of 6 dB followed by a second stage with 28 dB of gain with 6 positions of attenuation.  The result was a overall 60 dB gain range with approximately 6 dB steps.


The original gain structure is as follows:  (11 positions, 6dB gain steps)

Position  1 (-80dB) = +30dB Amp1 - 0dB pad +28dB Amp2 = +58dB total
Position  2 (-74dB) = +30dB Amp1 - 6dB pad +28dB Amp2 = +52dB total
Position  3 (-68dB) = +30dB Amp1 -12dB pad +28dB Amp2 = +46dB total
Position  4 (-62dB) = +30dB Amp1 -18dB pad +28dB Amp2 = +40dB total
Position  5 (-56dB) = +30dB Amp1 -24dB pad +28dB Amp2 = +34dB total
Position  6 (-50dB) = +28dB Amp2 - 0dB pad = +28dB total
Position  7 (-44dB) = +28dB Amp2 - 6dB pad = +22dB total
Position  8 (-38dB) = +28dB Amp2 -12dB pad = +16dB total
Position  9 (-32dB) = +28dB Amp2 -18dB pad = +10dB total
Position 10 (-26dB) = +28dB Amp2 -24dB pad = +4dB total
Position 11 (-20dB) = +28dB Amp2 -30dB pad = -2dB total


The following modified gain structure keeps the first stage amplifier at +30dB and the second stage +28dB.
This is one way to implement the "OFF" position with essentially no change to the 5 highest gain positions in which the first stage runs at +30 dB fixed gain and its output is attenuated and then fed to the +28dB fixed-gain 2nd stage.

Modified gain structure #1:

(5 high-gain positions, 6.0 dB steps; 1 "OFF" position;
5 low-gain positions, 7.5dB gain steps)

Position  1 (-80dB) = +30dB Amp1 - 0dB pad +28dB Amp2 = +58dB total
Position  2 (-74dB) = +30dB Amp1 - 6dB pad +28dB Amp2 = +52dB total
Position  3 (-68dB) = +30dB Amp1 -12dB pad +28dB Amp2 = +46dB total
Position  4 (-62dB) = +30dB Amp1 -18dB pad +28dB Amp2 = +40dB total
Position  5 (-56dB) = +30dB Amp1 -24dB pad +28dB Amp2 = +34dB total
Position  6 OFF
Position  7 (-50.0dB) = +28dB Amp2 - 0.0dB pad = +28.0dB total
Position  8 (-42.5dB) = +28dB Amp2 - 7.5dB pad = +20.5dB total
Position  9 (-35.0dB) = +28dB Amp2 -15.0dB pad = +13.0dB total
Position 10 (-27.5dB) = +28dB Amp2 -22.5dB pad =  +5.5dB total
Position 11 (-20.0dB) = +28dB Amp2 -30.0dB pad =  -2.0dB total


A second (IMHO better) approach to the gain structure would attempt to optimize the gain structure and also try to increase the input impedance of the first amplifier stage.  The greater the gain, the lower the input Z, and the 30 dB first stage sets the minimum load on the mic input transformer.

In this second scheme, the first stage gain would be reduced from +30 to +29 dB and the second stage would increase from +28dB to +29 dB, keeping the overall maximum gain the same as the original.
This significantly raises the first-stage input impedance that loads the transformer by lowering its gain. The 2nd stage input Z would be reduced, but both stages would be the same.


Modified Gain Structure #2:

(4 high-gain positions, 7.25dB gain steps; 1 "OFF" position;
6 low-gain positions, 6.2 dB gain steps)

This structure favors more switch positions going through only the second stage, bypassing the first stage altogether.
The high gain steps are more spread out and the low gain steps are closer together than the first modified gain stage proposal. The first low-gain position has 1 dB more gain than before.


Position  1 (-80.00dB) = +29dB Amp1 - 0.00dB pad +29dB Amp2 = +58.00dB total 
Position  2 (-72.75dB) = +29dB Amp1 - 7.25dB pad +29dB Amp2 = +50.75dB total
Position  3 (-65.50dB) = +29dB Amp1 -14.50dB pad +29dB Amp2 = +43.50dB total
Position  4 (-58.25dB) = +29dB Amp1 -18.00dB pad +29dB Amp2 = +36.25dB total
Position  5 OFF
Position  6 (-51.0dB) = +29dB Amp2 - 0.0dB pad = +29.0dB total
Position  7 (-44.8dB) = +29dB Amp2 - 6.2dB pad = +22.8dB total
Position  8 (-38.6dB) = +29dB Amp2 -12.4dB pad = +16.6dB total
Position  9 (-32.4dB) = +29dB Amp2 -18.6dB pad = +10.4dB total
Position 10 (-26.2dB) = +29dB Amp2 -24.8dB pad =  +4.2dB total
Position 11 (-20.0dB) = +29dB Amp2 -31.0dB pad =  -2.0dB total


Feedback is solicited on these two approaches.  Or maybe you have one that I haven't thought of!

 

[Steve Hogan]

Even More Info from Steve Hogan re: ACMP mic preamp cards
Amplifier Input Z vs gain

I spent several hours spice modeling the ACMP preamp in order to implement the gain changes from my last post and to fix some subsonic peaking issues which I will discuss in my next post.

Here are numbers for the input impedance discussion from my prior post.

The modelled mid-frequency (1kHz) input Z of the first stage amplifer at +30 dB of gain with factory feedback values = 3K59.

The modelled mid-frequency (1kHz) input Z of the second stage amplifer at +28 dB of gain with factory feedback values = 4K938.

The modelled mid-frequency (1kHz) input Z of both first and second stage with +29 dB of gain (modified feedback values) = 4K567.

The modelled mid-frequency (1kHz) input Z of the second stage amplifer at +18 dB of gain with factory feedback values = 8K645.

Note that by reducing the first stage gain by only 1 dB its input Z rises from 3K59 to 4K567. This is a 977 Ohm 27% increase in impedance and consequently lighter input transformer secondary load.  This very significant impedance increase is a strong vote in favor of the 4-position high-gain, OFF, 6 position low-gain configuration #2 discussed in the previous post.

 

[Paul G]

Hi Steve.  Thanks for posting.  From what I'm able to understand, I'd lean toward the second option.

Paul ;D

 

[Steve Hogan]

Mods from Steve Hogan to eliminate subsonic peaking in ACMP mic preamp cards

The following suggested modifications to the ACMP preamp circuitry solves some serious sub-sonic peaking of the factory frequency response.  The capacitor values in several locations have a very significant effect on the ultra LF response.  My recommended values will result in a smooth non-peaking flat response to 1Hz with a smooth roll-off below 1 Hz.  The factory values have greater than 3dB boost (>6dB through both stages) at 1 Hz and response that unnecessarily extends several octaves lower than 1 Hz. This peaking behavior results in extra pickup of subsonic garbage that will waste lots of subwoofer power and generate intermod as the audio rides on large amplitude subsonic crud.

These mods should not affect the desireable coloration of the preamp, but will eliminate and reduce some electrolytic capacitor ugliness in addition to fixing the ultra Low frequency peak.

Procedure:
Component references are from the '81 version of the PCB.
Refer to the factory supplied schematics:

1. Change 1st stage input cap 1C1(10uF/63V) to 100uF/25V
2. Change 2nd stage input cap 1C11(22uF/35V) to 100uF/25V
3. Change LF bypass caps 1C5 and 1C15 (220uF/25V) to 470uF/25V
4. Change Output caps 1C3 and 1C13(22uF/35V) to 120uF/25V or alternatively 100uF/25V
5. Remove 10K pull-down resistors 1R12 and 1R25 these must be removed in order to for this mod to work.  The feedback resistors serve the same function as the 10K pulldowns in the circuit.
6. Cut trace between 1C1 Input Cap and 1st stage feedback resistor 1R14 (15K factory (+30dB) or 18K2 if modified for +29dB gain). Jumper the cut away end of the resistor to the other side of 1C1.  This removes the input capacitor from the 1st stage feedback path.  The resistor is still AC coupled on the other end by 1C6.
7. Cut trace between 1C11 Input Cap and 2nd stage feedback resistor 1R27 (18K). Jumper the cut away end of the resistor to the other side of 1C11.  This removes the input capacitor from the 2nd stage feedback path.  The resistor is still AC coupled on the other end by 1C16.
8. Change 1R62(10K) to 100K to reduce loading on output cap 1C13 by the peak LED circuitry.

 

[tommypiper]

Hi Steve, thanks so much.  I've just read through the posts once quickly (I may need to digest them further later on) but my comments. I have modified an original Neve 1272 and boosted its gain to 70-something dB and used it successfully with an extremely low output RCA Jr Velocity ribbon mic, (about as low output a mic as you'll ever find -- way lower than modern ribbon mics or even lower still than the RCA 44 or Coles 4038).  I had no signal to noise issues whatsoever.  And that's a pretty unforgiving mic when it comes to noise...  just an anecdote here, that I don't think signal to noise will be an issue, if I'm understanding correctly. 

-Doesn't changing the input transformer load change the character of the amp?

-I don't understand the reason for wanting to change the loading, apart from aiding your scheme #2?

I would vote for scheme structure #1 because I work mostly with acoustic sources and low output mics.  Therefore I need the flexibility towards the high end of the gain spectrum. 

I also prefer the idea of using only one amp in the lower gain settings because, well, it seems better and it's closer to the original Neve design.  Therefore, again I vote for Structure #1. 

Third, I don't understand why we'd want to change the input transformer loading.  Again, the originals are great.  Please pursuade me otherwise. 

 

[Steve Hogan]

...I don't think signal to noise will be an issue, if I'm understanding correctly.  

I would vote for scheme structure #1 because I work mostly with acoustic sources and low output mics.  Therefore I need the flexibility towards the high end of the gain spectrum.  

Since I last posted, I have pretty much decided on gain scheme #2 which has 4 high-gain positions (-80 to -58.25dB) with 7.25dB steps, "OFF", and 6 low-gain positions (-51 to -20dB) with 6.2dB steps.  Amp1 gain goes from +30dB to +29dB, and Amp2 goes from +28dB to +29dB.  The final (sheme #2) modified gain structure compared to the original is as follows:

Position 1   Was -80dB >now -80dB       (2 gain stages)
Position 2   Was -74dB >now -72.75dB   (2 gain stages)
Position 3   Was -68dB >now -65.5dB    (2 gain stages)
Position 4   Was -62dB >now -58.25dB  (2 gain stages)
Position 5   Was -56dB >now "OFF"       (was 2 gain stages > now "OFF")
Position 6   Was -50dB >now -51dB      (1 gain stage)
Position 7   Was -44dB >now -44.8dB    (1 gain stage)
Position 8   Was -38dB >now -36.6dB    (1 gain stage)
Position 9   Was -32dB >now -32.4dB    (1 gain stage)
Position 10 Was -26dB >now -26.2dB    (1 gain stage)
Position 11 Was -20dB >now -20dB      (1 gain stage)

It is true that the above scheme sacrifices step size on the high-gain end, but the smaller step size of the original ACMP and my revised gain scheme #1 just allows more padding after the first +30db of fixed gain. My proposed scheme uses more of the +30dB amplifier when needed.
It also allows you to bypass the first stage completely 1 dB sooner and spreads out the gain steps more evenly
(7.25+6.2dB steps instead of 6.0+7.5dB steps).

I am now implementing this modification to one of the '81 preamp cards.  I don't have all the resistor values required, but I hope to still check it out this weekend. The pop can be eliminated (I think) without having to cut any traces. One (new value) resistor must be inserted diagonally across the space of 2 previously installed resistors.

The original Neve circuits had more poles on the gain switch and many more positions than the ACMP.
One of the poles was used to change the gains of the 2 amplifiers so that the gain setting was a combination of feedback gain control and pads.  The ACMP uses only a single fixed gain for each amplifier when in mic mode.
The ACMP preamps use clones of the amplifiers -- same topology as the Neve, but the gain structure is quite different, so we have to adapt to squeeze the best performance.

 

I also prefer the idea of using only one amp in the lower gain settings because, well, it seems better and it's closer to the original Neve design.  Therefore, again I vote for Structure #1.

Both the original ACMP circuitry and both of my proposed modified gain structure schemes use only one amp in the lower gain settings.  You may have misunderstood that aspect of my proposed fix.

-Doesn't changing the input transformer load change the character of the amp?

-I don't understand the reason for wanting to change the loading, apart from aiding your scheme #2?

I don't understand why we'd want to change the input transformer loading.  Again, the originals are great.  Please pursuade me otherwise.  

In short, optimizing the loading on both the line and mic input transformers will change the character of the amp -- for the better.  Remember that the Chinese just slavishly copied various bits and pieces of Neve circuitry while using different components. Remember the '81 circuitry where they substituted transistors with too much gain and got an oscillation and they substituted a 1N914 silicon diode instead of the Neve germanium for the Baker clamp.  They keep 22uF caps as in the original Neve, but they loaded them down with peak LED circuitry which changed the Low end.

At this point, I will lean on my over 20 years of experience as VP Engineering at Jensen Transformers, where I designed the entire JT-series product line.  I cannot remember a case where overshoot and ringing of the input transformers sounded better than that same transformer properly loaded with the proper RC network.  A ringing tranformer rings like a bell with every steep transient and its contribution has no harmonic relationship to the music.  The resulting intermod contributes to a glaring smeared midrange.   Somewhere in this thread there is a contribution from someone who reworked the transformer loading already with good effect. Not every transformer can be properly damped, some have multiple resonances that make it ring no matter what you do.  Those never made it at Jensen.

The good news is that both the Line and Mic input tranformers can be damped quite nicely.  As supplied by the factory, however, their ringing is made much worse by a capacitor placed directly across the secondary.  This makes the ringing much larger and lower frequency than if the capacitor wasn't there.  I plan to add a proper RC network to both transformers in the units I modify here.

The mic input transformer is a 1:2 ratio when in normal mode, and a 1:4 ratio when the switch in the back is pressed.
The resistive loading on the transformer is reflected to the primary by the square of the turns ratio. Thus, (ignoring the DCR of the windings for the moment) the microphone will be loaded by 1/4 the resistive load on the secondary in 1:2 mode and 1/16 the resistive load on the secondary in 1:4 mode.

Depending on the type of microphone, the load presented to a microphone can have either a major or minor effect on its output level and its frequency response.  It can also effect the damping of a ribbon and the voice coil in a dynamic mic.  Loading it down damps it more, causes the mic to lose level, especially ribbon mics with high output Z.

To maximize voltage transfer and thus S/N ratio, it is best to bridge microphones with a load about 10 times their output Z, so for a nominal 150 Ohm microphone, a load of 1500 Ohms is about right.  The Jensen JT-16-A 1:2 ratio mic input transformer uses a load resistor of 6K19 on its secondary, thus the nominal load presented to the mic is 6K19/4=1547.5 Ohms. Because of the DCR of the windings, the load is actually a bit higher, but you get the idea.

With 3K59 input Z from the first stage in parallel with the 12K resistor in the factory circuit, The ACMP's input Z is 2663/4= 691 Ohms in 1:2 and 166 Ohms in 1:4.  For this reason I strongly recommend against ever using the high gain button on the back of the preamp as it will severely load any microphone plugged into it.  In the case of ribbon mics that can have output impedance as high as 600 Ohms, the loading will cause a severe loss in level, and a loss of High frequency bandwidth.

In the original ACMP circuit the input impedance changes depending on the gain, so that the load on the mic is not constant.  Like the incorrectly designed mic pads that change the character of the sound instead of just changing the level.

In my reworked gain structure I have been able to provide a constant resistive load to the mic transformer of 4563 Ohms +/-about 25 Ohms and still keep all the gain steps precise to less than 0.1dB.  The resulting load to the microphone is 4563/4 = 1141 Ohms which is a lot better than 691.  Add to that a new RC network to eliminate the overshoot and ringing and you will have a mic preamp that will sound better than the original by a lot.

Hopefully, you will be persuaded

 

[tommypiper]

Thanks.  Yes, I'm aware of all that loading theory, but it was not on the tip of my brain.  Thanks for going over it so thoroughly.

I have a well regarded transformer expert friend who says never worry about transformer ringing.  (On outputs at least.  I don't know if he feels that way about input transformers as well.)  There seem to be two schools of thought on the subject.  I know it's been discussed here in the forum... we don't need to get into that though...

Looking forward to your RC network.

 

[rodabod]

Tom, I found the standard transformers sounded a little harder than the Carnhills as stock. I then added a Zoebel to the original input and output which I measured out myself using the standard Jensen procedure, before finally swapping them out completely.

I found that 600R loading on the output had an impact, but I'm not going to say what (it was measurable and listenable, and people can try this themselves).

 

[maxwall]

Regarding the ACMP81,

I find proper impedance loading is significant thru the gain range and realize its importance in keeping the loading as constant as possible, afterall this is part of the design concept behind the Neve loading as well. Why not address two issues at once, the "pop" and the impedance loading all in the same fix. A wise choice.

I can live with the gain spead due to new resistor values which is minor in comparison to the two aformentioned issues. Why , because I can always use the variable gain pot , not the rotary gain switch, to fine adjust exactly where I want the gain to be. This is the variable output pad pot on the far right hand side of the ACMP81 opposite the rotary gain switch labeled "output".

I opt to choose gain structure #2

Keep up the fine work Steve !!!

 

[Steve Hogan]

Quick update from Steve Hogan

I installed all the calculated resistor values on the gain switch to implement the Gain structure #2 from the previous post, and have done some preliminary examination of the results. The gain steps came out exactly as predicted (+/-0.1dB). The pop is gone, and I added a 100 Ohm resistor to ground for the "OFF" position which makes the "OFF" position dead quiet. This mod does not require trace cuts, just new resistor values and a couple of them inserted "crooked".

Finding an optimal RC Network, however, looks like it might be a little bit tricky.  When the secondary of the mic input transformer is loaded directly by the input of one of the amplifiers, (all the high-gain inputs, and the first low-gain step) there is a significant capacitive load on the secondary caused by the 1500 pF capacitor between base and emitter of the first transistor in each amplifer.  The remaining low-gain steps provide a resistive pad load to the transformer secondary which isolates it from the capacitive amplifier 2 input.

Capacitance across the secondary along with the leakage inductance in the transformer causes a resonance that makes the transformer ring.  The RC network adds additional secondary loading at high frequencies to damp that resonance.  When the amplifier input is (very) capacitive, it peaks the transformer's magnitude response, so a lower impedance RC network is required to straighten out the transient response than what is required for a resistively-loaded transformer secondary.

The preliminary RC network required to damp the mic input transformer on the high range is more aggressive than the one needed at low-gain settings.  I think that I may actually try 2 networks -- One that lives permanently across the transformer secondary to damp the low-gain, resistively loaded steps, and a second one that is added only when the transformer secondary is directly connected to the amplifer input.  Maybe the two networks in tandem will work better (give higher bandwidth and faster rise time) than a single, more aggressive one.

I will keep you posted as I find out more.

 

[Rockit88]

........followed by months of silence............again

 

[tommypiper]

Finding an optimal RC Network, however, looks like it might be a little bit tricky.  When the secondary of the mic input transformer is loaded directly by the input of one of the amplifiers, (all the high-gain inputs, and the first low-gain step) there is a significant capacitive load on the secondary caused by the 1500 pF capacitor between base and emitter of the first transistor in each amplifer.  The remaining low-gain steps provide a resistive pad load to the transformer secondary which isolates it from the capacitive amplifier 2 input.

Capacitance across the secondary along with the leakage inductance in the transformer causes a resonance that makes the transformer ring.  The RC network adds additional secondary loading at high frequencies to damp that resonance.  When the amplifier input is (very) capacitive, it peaks the transformer's magnitude response, so a lower impedance RC network is required to straighten out the transient response than what is required for a resistively-loaded transformer secondary.

The preliminary RC network required to damp the mic input transformer on the high range is more aggressive than the one needed at low-gain settings.  I think that I may actually try 2 networks -- One that lives permanently across the transformer secondary to damp the low-gain, resistively loaded steps, and a second one that is added only when the transformer secondary is directly connected to the amplifer input.  Maybe the two networks in tandem will work better (give higher bandwidth and faster rise time) than a single, more aggressive one.

I will keep you posted as I find out more.

Makes one appreciate the simplicity of Rupert Neve's approach: adjust by ear, done.

As I said at the top of the page, there is a school of equally educated people who don't buy into all this RC secondary loading stuff and instead simply let transformers ring, etc.  Why?  Because they still sound fine, and in many cases better! (Again, I know that applies to output transformers, not sure about inputs.)

Footnote: just to be fair, RN was of course aware of the issues of loading and such and did not ignore it, I don't mean to oversimplify, as he loaded outputs. But it's been said many times his final analysis was not by theory, but by ear.

 

[tommypiper]

I found the standard transformers sounded a little harder than the Carnhills as stock. I then added a Zoebel to the original input and output which I measured out myself using the standard Jensen procedure, before finally swapping them out completely.

I found that 600R loading on the output had an impact, but I'm not going to say what (it was measurable and listenable, and people can try this themselves).

Thanks.  Yep, my transformer friend says the 600R loadings "always suck the life out of the sound."  If/when I put Zoebels and loads on I plan to thoroughly test and evaluate by ear, and when it's useful have it switchable in and out. 

Interesting that you found the Jensen procedure Zoebel of no use. 

My friend swears that transformer ringing is of absolutely no audio consequence, at least in the majority of cases.  "Let em ring," he says.  I'm sure he also believes using the correct transformer will make a difference.  But if it happens to ring, let it ring, is his view.

 

[crazydoc]

I will keep you posted as I find out more.

So, Steve - it's been 6 weeks now. (Probably just a split second in your geologic time frame of reference.  ;D)

 

[willowhaus]

........followed by months of silence............again

er...no.

Steve Hogan
Fullerton, California

Re: [ACMP investiupgradifications] All things PREAMP
« Reply #963 on: March 22, 2010, 09:11:43 pm »

:

 

[tommypiper]

Yeah, this is taking wayyy too long. 

As I've been trying to hint at on this page, I feel the ultra-überanalysis of an RC damping factor for the transformers is probably overkill.  I'm interested in Steve's result, I respect the process and the drive to perfection.  But the RC damping transformer network obsession looks like an over-the-top analysis and a super complicated approach which is probably out of proportion for this project and is taking waaay too long.  This is just a $200 pre, for crying out loud. 

See Rod's post after mine at the top of the page.  He went through a similar analysis for treating the transformers and found it of no use! 

Come on Steve, let's move on!

The work you've done so far has been helpful, thanks, but let's not obsess over a little transformer scope reading at 5 gazillion megahertz.  My ears will tell me what's good or bad, I don't want to complicate this fix more than necessary.  If it's too complicated, please just recommend another transformer, like  Lundahl.  (Lundahl make Neve, gapped output clones now.) 

 

[rodabod]

Rod's post after mine at the top of the page.  He went through a similar analysis for treating the transformers and found it of no use! 

I should probably clarify that what really happened was, I "optimised" some Zoebel values for the original transformer, but ended up scrapping the transformers with their new Zoebels completely, in favour of some new Carnhills. I'm not saying that they were bad. I have many measurements and also recordings compaaring the performance of the transformers. The output models actually spec quite close to the Carnhills in some respects.

I did what anyone with reasonable DIY skills can do; swapped a few parts out, and twisted the mains transformer. It's not tricky. My measurements were timely and required some knowledge, but it certainly wasn't rocket science.

Steve, you probably shouldn't have taken this on if you weren't able to provide a prompt service. This may be a DIY forum, but when you charge money, people expect a professional service. It wasn't a favour from what I can see, unless you were charging break-even prices. I don't know how many you need to provide for, but it's at least polite to keep people up-to-date, let alone give them what they paid for!

 

[mitsos]

I have used mine with a new zobel for a couple of things, and it sounds fine. As Roddy says, swapping the trafos is not necessary (for me, it was a decision mainly because carnhills are difficult/expensive to get here).  I'm sure Steve can/will find many shortcomings of the design/layout, and that may make these preamps quite a bit better than they originally were.  I can also understand how these things take so long, after trying for ages to get a simple preamp PCB done (revision after revision after revision).

But I have to agree with Roddy, there is a line crossed when you take money from people and then months(?) later there is nothing delivered.

That said, I have a lot of respect for Steve's technical ability (I wish I could say I worked with Deanne Jensen) and he seems like such a nice guy, but DUDE, just type something up here. Even if it's "hi guys, been really busy, my daughter's had the flu and I haven't had a chance to do anything."  Whatever, just let people know. 

please and thank you. 

 

[okgb]

Hey Roddy , can you recap
what worked best for you with the original xfmrs and
what worked best swapping them out ?
thanks & regards Greg

 

[rodabod]

Hi Greg,

I've emailed you my marked-up schematic (possibly again).

Basically, it was fine as it was with the toroid rotated. I then listened to an A/B of two very similar Chinese copies with the original Chinese iron, and the other with Carnhills. Basically, the Carnhills seemed to have a slightly thicker bottom-end and low mids, whereas the Chinese models either sounded harder or thinner (maybe they were just cleaner, who knows?).

So, I then spent about a week making measurements and adjusting values, including new Zoebel networks to damp the Chinese mic input transformers for best square wave / amplitude response. The originals peak at the top end of the audio band. Not massively, but they do. If I remember correctly, the output transformer is over-damped, even without 600R termination.

That said, I didn't notice a massive improvement with the newly added networks. In one case, I found that it sounded worse and had to compensate for what I had originally calculated/designed. You should also be aware of mic loading; that secondary impedance gets stepped-down on the mic transformer...

With the new transformers, I also replaced a large number of caps, where these might affect audio signal quality.

Hope that helps.