home | a note | power scaling report | vintage marshall chassis pics

bc audio | my youtube channel
 

Power Scaling Report
Bruce Clement, January 2003

Power Scaling is a technology developed by London Power, headed by renowned tube amp innovator and author Kevin O'Connor, that controls voltages inside the amplifier, allowing for continuously adjustable power output.  Power Scaling reduces power, and thus volume, enabling the amp to produce power tube distortion at reduced volume levels, freeing you from the need for an attenuator.  It consists of two solid state voltage regulators - one for B+, one for bias - controlled by a single pot for continuous adjustment from near 100% power down to less than 1%.  Additionally, four bias pots are provided for individual tube bias.  For more info, read these pages on the manufacturer's web site:

Power Scaling Kits | Power Scaling Q&A | Power Scaling Kit FAQ

My goal in this project was simple - get the amazing cranked tone of my 100 watt Marshall Plexi Reissue, at lower sound pressure levels, without the use of an attenuator.  In addition to the Power Scaling Kit, this amp has been modified with an original 1968 plexi output transformer and a point-to-point eyelet board, following an early JTM100 circuit.


The subject amp and the speaker cabinet used for recording Tone Samples.


PSK board mounted at power supply end of chassis.  Note four board-mounted pots for individual tube bias and four 1-ohm bias sensing cathode resistors.  Power Scale pot is at lower left, mounted where Presence used to be.  MOSFET is mounted to heatsink below fuse holders.


Another shot of the board. 


Fan mounted in filter cap hole.  Heat sink is just below.


Heat sink for MOSFET.  Note nicely braided wire; my wife helped me with that.


Close-up of heat sink.  There's more clearance between the heat sink bolt and the fuse holder than there appears.


Now this would be hokey if it weren't so clever.  This is the fan power supply.  It's the guts of one of those "universal" wall-wart adaptors that can switch between 3, 6, 9 and 12VDC.  I have it set so the fan is running below its rated voltage so it's quiet but still moves air.  The white stuff on the chassis is what's left of sticky-back cable ties installed by a previous owner for a since-removed mod.

Overview

First let me say that this was not a simple weekend project!  Installing this kit required that I fully understood my amp's power supply design, knew my way around the amp's guts, was able to really read a schematic and board layout, and had good electronics construction habits.  If you've never been inside an amp before, don't try this at home!

Also to be considered was that the kit would add two new controls to the amp - Power Scale and Drive Compensation.  For me, this meant moving Presence and Bass to the rear panel so I could have the PSK controls on the front of the amp.  If it were just one control, I would put Presence in one speaker jack hole, but since there were two, I would have had to - gulp - drill.  In my case, there were already holes in the rear of the chassis from a previous owner's effects loop, which I had removed.  Drilling would not be acceptable on a vintage amp!

The Kit

The Power Scaling Kit (PSK-1, US$148) came as parts in Ziplok bags plus several pages of photocopied hand-written instructions.  Parts included a MOSFET, a couple of small transistors, several diodes, resistors and capacitors, four small bias pots and a bigger 2w pot for the Power Scale control (my kit was an early one, and did not include a Drive Compensation control - more on that below).

I also bought a Circuit Card Kit (CCK, US$42) for making up the eyelet board.  The CCK included enough board material and eyelets to make an additional board to replace the amp's stock printed circuit board, which I did after I had completed the PSK.

Construction

A board layout was provided in the PSK instructions, but I came up with my own because I wanted to mount the four bias pots and cathode (bias sensing) resistors on the board.  I also elected to put the rectification on the board.  I came up with a good layout on a 3"x6" board which fit at the power transformer end of the chassis, mounted on short standoffs.

I drilled the board, installed the eyelets and soldered in the parts.  I then installed the board in the amp, connected it to the power supply and output section and mounted the Power Scale pot.  I tested voltages without tubes, made some tweaks to get the bias voltage in the right range, and I was ready to go.

I put in my tubes and fired 'er up.  Set bias, no problem.  I found that Power Scale adjusts B+ from about 450VDC down to less than 20(!).  B+ was 465 stock, so there is some loss in the regulator circuit, which I anticipated, and is not a problem for me, though it reduces maximum power somewhat.

Bias voltage tracks with B+.  The four bias controls have a very wide range, much more than a typical bias adjustment pot, so any tubes can be installed and biased right up without changing any resistors.  Because of the four individual controls, non-matched sets or even different tube types can be used, according to London Power.

At this point, I ran into two problems.

Problem #1: Increased Power Amp Sensitivity

I had expected that as I turned down the Power Scale control, the output signal would basically shrink, but not otherwise change. So, if I was playing with a certain amount of distortion, when I turned down Power Scale, I would still have the same amount of distortion, only quieter. Not so. As Power Scale is turned down, the amount of signal required to drive the power section is reduced, too.  Without the Drive Compensation control, distortion increases dramatically as the power amp's clipping point moves downward, until even the smallest signal drives the amp into clipping.  You can imagine what this sounds like.  As you turn down Power Scale, the amp's maximum power goes down, but it also gets much more distorted, and when you stop playing, the noise isn't any quieter.

Remember, we're talking about a non-master-volume amp, in which there was no way to reduce the signal being fed into the power section.  Turning down the amp's volume control helped some, but changed the tone, because the preamp was no longer being driven hard, so there was no preamp distortion in the mix.

Of course, this was not the result I expected.  Several emails to Kevin O'Connor resolved the problem and, I believe, resulted in the Drive Compensation control becoming a stock part of the kit.  Drive Compensation is electronically similar to a Master Volume, but as it is being used here, it functions quite differently.  Normally, a MV isolates the preamp from the power amp so you can overdrive the preamp and send some of the signal to the power amp.  Used with the PSK, it compensates for the reduced drive requirements of the scaled-down power amp, preventing excessive power amp distortion.

Drive Compensation lets you maintain the correct mix of preamp and power amp overdrive at any volume level as the power amp is effectively resized via Power Scale.  While it is somewhat of a pain to dial in two controls, it works quite well.  An ideal solution would be to electronically (or mechanically) track the two controls together with one knob, maintaining a correct amount of drive at all times.

Problem #2: Excessive Heat

I had a lot of problems with heat.  As I was beginning to test the installed kit, I found the MOSFET would get very hot - hot enough to burn my finger instantly.  The PSK came with two small aluminum heat sinks for the MOSFET.  Alternatively, the instructions indicated that the MOSFET could be mounted directly to the chassis.  I was using the heat sinks and did not try the chassis-mounting idea.

The kit was doing what is was supposed to do, but the heat became too much, and the MOSFET blew.  Kevin was kind enough to send me a new one, free of charge.  Before I toasted the new one, I knew I had to come up with much more heat dissipation.

I devised a solution involving a small computer CPU fan and heat sink.  The fan sits on top of the chassis directly over an existing filter capacitor hole (I moved the cap under the chassis), with the heat sink directly below.  Screws hold the fan and sink together, sandwiching the chassis between them (see pics).  The 12VDC fan is powered by an adjustable-voltage A/C adapter mounted inside the chassis.

This cooling scheme works quite well, and I've had no problems with heat since installing it.

UPDATE:

Still not satisfied with the heat issue, I wanted to try a different approach.  I purchased London Power's Cascode kit ("PSC," apparently no longer listed on the London Power web site).  This kit included another MOSFET, discrete parts and instructions.  With two MOSFETs to share the load, heat is greatly reduced.

I mounted both MOSFETS to the end of the steel chassis with mica insulators and heat sink goop.  I removed the fan and fan power supply, and I put he cap back in the hole where it belongs. 

The MOSFETS and chassis get nice and warm, but never as hot as before.  I've gigged with the amp several times, rehearsed with it a lot, and I've never had a problem with the PSK+PSC.

What to Scale?

The PSK's instructions tell you that it is possible to scale just the power tubes (leaving the preamp section to run on normal voltages), or to scale the whole amp.

I found that scaling the whole amp (power amp and preamp) changes the sound and feel of the preamp as the bias on the 12AX7s change, and gain is reduced significantly.  This clearly wouldn't do.

Scaling the power tubes alone was definitely the way to go, but there is one flaw in the suggested approach.  Drive Compensation is designed to go after the phase inverter (PI).  The negative feedback (NFB) loop returns to the input of the PI, before the Drive Compensation control.  So, with lower settings of Drive Compensation, less NFB is making it back to the PI and power amp, resulting in the amp's overall gain - and graininess - increasing, and the Presence control becoming less effective.

I solved this problem by moving Drive Compensation to before the PI, as in a traditional Master Volume after the tone stack.  This way, the NFB loop signal is not reduced by Drive Compensation, the open loop gain of the amp is unchanged, and Presence works normally.

Operation and Tone

So, how does it sound?  I think it sounds like the same amp, only quieter.  I own a Marshall Power Brake, a THD Hot Plate and a couple of home-brew resistance-only attenuators.  At any given Power Scale setting, the amp sounds much more natural than it ever did at the equivalent volume with any attenuator.

The great thing about the PSK is this: I can now actually enjoy the cranked tone of my amp.  I rarely - and I do mean rarely - had the opportunity to crack my 100w Marshall wide open without an attenuator, so the tone to which I was accustomed was what came out of an attenuator, not the true tone of the amp.  With the PSK, there's nothing between the amp and the speaker to change the tone.

Now, we can't ignore issues like speakers not being pushed as hard, the non-linear response of the human ear, lack of acoustical feedback to the guitar, and whatever else you want to throw into the mix.  Ultimately, a quiet amp will never sound the same as a loud amp.  But I think the PSK does a good job of doing exactly what it is designed to do - let the player scale the amp to the situation, while maintaining its true tone.

Conclusions

London Power's approach to the problem of too much volume is truly innovative.  But it is not a perfect solution.  It requires significant modification to the amp, including chassis drilling; it's a lot of work to install, and it should only be done by a qualified amp tech or a very experienced hobbyist; it cannot be moved from one amp to another as attenuators can; it should not be installed in a vintage or otherwise expensive amp.

There are many ways that people try to get good power tube tone at lower volumes, from the simple ("get a Hot Plate") to the cumbersome (multi-amp slaved rigs, isolation cabs) to the expensive (a different amp for every possible venue).  The PSK is, in my opinion, a very effective but not very practical solution.  I would never install one in a vintage amp, nor would I put one in every non-collectable amp I own because of the sheer amount of work involved.

Having said that, I am delighted with the performance and tone of the amp with the PSK.  You might not expect a 100 watt Marshall running on one-tenth its normal voltage to sound anything like its normal self, but it does.

The PSK has a few other advantages over attenuators.  It is infinitely adjustable - as apposed to 3 or 4dB steps - and it's contained within the amp itself, making it easy to transport.  What's more, the four individual bias adjustment pots and the wide range of bias voltage available are a vast improvement over the stock bias circuit.

With its radical power supply hidden under the hood, this amp is not exactly a faithful replica of a vintage design, at least not in the power supply department.  Yet it arguably does a better job of producing the tone of a vintage 100 watt Marshall at reasonable volume levels than it ever did as a stock amp in combination with an attenuator.

But don't take my word for it.  Check out the Tone Samples (below) and judge for yourself

The Bottom Line - Comparison to Attenuators

The Tone Samples in the table below were recorded in order to demonstrate the effectiveness of the PSK compared to two popular attenuators.  I chose the THD Hot Plate and Marshall Power Brake for very specific reasons: they are the two most widely used attenuators on the market, and... that's what I happen to own.

My goal here is not to convince you that the PSK is superior to any particular attenuator.  My purpose is simply to provide objective data from which you are free to draw your own conclusions.

These sound clips were recorded by me in a make-shift project studio in my rehearsal facility, not in a anechoic chamber under laboratory conditions.  However, they were all recorded in the same space, in the same way, on the same day, with the same equipment, so they do offer a solid basis for fair comparison.

Each Tone Sample consists of a simple riff, played twice, with lots of ringing open strings and not too much gain.  There's enough distortion to give it good harmonic content, but not so much that the sound of the guitar is obliterated.

Tone Sample Table

Actual SPL1 PSK Voltage2 THD
Hot Plate3
Bright & Deep On
THD
Hot Plate3
 Bright & Deep Off
Marshall
Power Brake3
123dB(A) Stock tone, Power Scale and Drive Compensation on 10, no attenuator
121 380VDC -4dB - -
119 250 -8 -4dB -3dB
114 160  -12 -8 -9
110 97  -16 -12 -12
106 60  -16/84 -16 -15
101 34  -16/54 -16/74 -21
92 17  -16/24 -16/2.54 -30

Notes:
1 Maximum Sound Pressure Level in decibels at 1 meter, A-weighted.  Measurement taken on a Radio Shack model 330-2050 SPL Meter.
2 B+ Plate Voltage in Volts DC, measured at idle (no signal).  Reading taken at output of PSK (1st filter cap/input side of choke).  Tone Samples in this column were recorded with no attenuator.
3Attenuator dial setting, in decibels below full volume.  Because of variations between attenuators, and in the case of the Hot Plate, variation between tone switch settings, and because what a manufacturer calls 3 or 4 dB is approximate, this figure may not correspond to the actual dB reduction.  Tone Samples in these columns were recorded with Power Scale and Drive Compensation controls on 10.  For the Hot Plate, the bulb switch was off.
4The second figure is the setting of the Hot Plate's fine adjustment control, on a scale of 0 to 10.

How to Read - and Listen to - the Tone Sample Table

In order to make fair comparisons, we have to hear each device at the same volume as the others, hence the reference to actual sound pressure level (SPL).  Each yellow "Play" button plays an audio recording (mp3 file).  To compare one device to another, listen to all of the Tone Samples in a given row.

To hear how a particular device sounds at different volume levels, start with the "Stock Tone" recording a the top, then listen to each of the samples in any one column, moving down the column to hear how the tone changes - or doesn't change - as actual volume is dialed down.

Please note that all Tone Samples are equal volume when played back through your speakers or headphones; for each recording, the input level of the mic preamp was adjusted to obtain a suitable recording.

Recording Notes

For these recordings, a Gibson Les Paul was plugged into the Marshall's Channel I high input only, no bridging.  Guitar Volume and Tone were set to 10.  Amp control settings were as follows:  Presence - 0, Bass - 5, Middle - 5, Treble - 5, Volume I - 6, Volume II - 0.  The speaker cabinet was a Marshall reissue 4x12 slant with Celestion Vintage 30 speakers.  Mic was Shure SM57.  Pic pre was a JOEMEEK VC3Q (minimal compression, EQ bypassed).  Recorded on a Roland VS-840GX, with a little bit of onboard reverb added.

2003 Bruce Edward Clement