Precision Power Scale Circuit

[K O'Connor]

Hi Guys

In the quest towards a Power Scale circuit with "ideal" control, I developed some alternative circuits, some similar to the current kits and others that are quite different.

As discussed in the "Power Scale pot sweep" thread, the nominal issues are to avoid having dead spots in the sweep at the ends of the pot rotation, and to have reasonable resolution of control at the quiet end. One would suppose that the ideal solution would be to use an opamp with a tailored response to potentially allow the use of a common llinear pot but achieve a modified log control shape? Opamps are wonderful devices and there are lots of low-power types available. However, they introduce their own problems and would still require interfacing to the high-voltage circuitry. Feedback would definitely be incorporated, and would have to be fancy to achieve the response mentioned, as well as be stable over the entire range of input voltages. The kit afterall has to accommodate the whole range of supply voltages of all the guitar and bass amps on the market.

Of course, it is simple enough to build a discrete high-voltage opamp with flexible input voltage capability, and that is one of the paths investigated. Over a supply voltage range of 200V up to 880V it worked perfectly after some refinements were added. In this design, both NPN and PNP transistors are used and even from the same manufacturer they are not available with the same voltage capabilities. Placing devices in series is a necessity - just like the current kits have - but three devices in cascode for the NPNs and two devices for the PNPs. This is only a minor inconvenience. For precision, all of the resistors are metal-film 1% 600mW types wired in parallel or series to achieve the required net values and to accommodate the heat dissipation at the highest input voltage. All of this adds up to a lot of components, so a "sandwich" board assembly is required, that is two PCBs that plug together and are parallel to each other.

Overall, the high-voltage opamp approach can be made quite small. I laid out the first version using cordwood style resistor installation. The solder pads were too small and too tightly spaced and I could not see myself soldering it, nor the average hobbyist. Someone with fantastic soldering skills and/or a magnifier, and/or maybe someone who does some surface-mount soldering would have no problem. Anyway, I laid out a second then third version to make it more "buildable", which has all the Rs lying down (SVn-TH, not built or released).

One of the amendments added to the circuit allowed "programmable" output voltage at the quiet end of the pot sweep. Four DIP switches allow this selection. There was a 4-way rotary switch alternative that turned out to be very low-quality. Overall, the control board began as slightly narrower than the present SV1 board width of 2.3" (59mm) and ended up slightly larger at 2.6" (66mm). The power board changes for SV1 and SV2 applications. The two boards plug together with 0.1" header pins and receptacles, then two bolts with lock nuts to secure them safely.

I found that the heart of the programmable amendment could be added to the present kit circuit and that this greatly enhanced the sweep of the Power Scale pot. The minimum voltage is set to a fixed value and the pot sweep is very good. The board size for the SV1 increased by 0.15" (4mm) width and the SV2 remained at its present width. This keeps everything familiar for techs who have installed a lot of the current kits. These new versions are still called SV1 and SV2 as common designators, but have a smaller designation of SV1-M36 and SV2-M36. The 'M' designates a change to 1M pots for Power Scale and Drive Compensation, and the '36' refers to the 36k resistor in parallel with the PS pot. These boards arrive before the Holidays.

One other amendment to the new kits is the use of 3W resistors in some positions. These are marked on the schematic and the PCB as there are still two positions using 1W of the same value. As always, it is best to sort all the parts before assembly.

Have fun

[K O'Connor]

Precision Power Scale part-deux

Hi Guys

Surprisingly, the first two Power Scale kit circuit forms we released were "precision" circuits. The basic circuit used the mil-spec pot that could sustain the B+ of most guitar and bass amps, and this controlled the mosfet directly. At the top end of the pot sweep there is just the Vgs loss of the mosfet; at the bottom end there is a non-zero point to keep the mosfet conducting. Again, the mosfet output is Vgs lower than the pot wiper voltage and could go to zero with random mosfet samples.

The first form is what we now refer to as "Classic-PS", which was copied by many. The second form is the "Super Budget" type (SB), described in TUT4 and TUT6, and was the first release of DC Power Scaling kits. Many techs and amp builders still use this and we supply PCBs for them.

The SD form used in my amps for many years was tuned by ear with respect to how the pot sweep works at the quiet end. For my particular power amp circuit, the point where the audio goes to zero is not where the supply goes to zero, and from recent feedback from installers, it is higher than for more common amp circuits. This latter detail means that the programmable "quiet" voltages for the circuits described in post-1 may be too high for many players, and that the idealised goal of achieving zero output voltage is not without merit.

With the kits of the recent years, there is a compromise of "where to put the dead spot" in the pot sweep. In some installations there will be no dead spot because the voltages in the amp and all of the parameter tolerances in the SVn, SVn-D and SVn-M36) circuit achieve a synchronicity. Otherwise, there may be a dead spot at the loud end or the quiet end of the pot sweep. The top end dead spot is only of concern for players who typically run their amp flat out and want to turn it down just a little bit. The lack of loudness change for the first part of the pot sweep is then disappointing. For most players who are trying to achieve bedroom SPLs, a dead spot at the quiet end is more of an issue.

If you have an issue with an SV installation, let me know and we can optimise some of the circuit values for the supply voltage in your amp. Or let us know when you order.

The latest circuit I have has no dead spots (SVn-Zh). As soon as the pot is moved from full-CW the voltage is dropping and it gets to zero exactly at the full-CCW position. For installers who think the present kit circuit is wonky, they would undoubtedly find this new one to be totally bizarre

[K O'Connor]

Precision Power Scaling part-C

Hi Guys

When I'm laying out the PCBs for SV1 and SV2, I do SV1 first then remove the parts not needed for SV2. Usually I can pull back one edge of the board so the empty space on SV2 is not too large. However, with the latest design (unreleased, but boards ordered and arriving soon) there was enough space left to add VCK, which is required with SV2 anyway.

Back when I began offering Power Scale kits on PCBs, some of the boards were GIGANTIC. In retrospect I do not know how these were fitted into a lot of amps? In the case of the kits for cathode-biased amps, an earlier form of VCK was part of the board. So, in that regard, the new board has come full circle but in a much smaller package.

Life is often circular and you hope that you learn something each time around.

[K O'Connor]

Hi Guys

It seems I was too enthusiastic with my board ordering - haha.

The SVn-M36 boards worked okay but I forgot a trace for the bias regulator on SV1-M36, which caused some inconvenience to a couple of installers. While those were being sorted out, I had already ordered boards for the SVn-Zh described above. The Zh has an ideal sweep from full B+ down to zero regardless of the B+ value. Its design range for input voltage is 200V up to 700V, so it accommodates every guitar or bass amp on the market.

I was wrong in my statement that Zh has no dead spots. The fact the output voltage can go to zero volts means there is a dead spot of the Power Scale pot sweep at the quiet end. Between the arrival of the M36 boards and of the Zh boards, I had feedback from some installers that lead to a modification of the Zh to the Zp - hopefully the ultimate in Precision Power Scaling.

The SVn-Zp has the same sweep of voltage that the Zh has, BUT there is a built-in user adjustment to fine tune the "audible transition" from silence to sound and vice versa. The AT trimmer lets the player set the loudness at the PS=0 end of the Power Scale panel pot. So now we have a Power Scale circuit that is not voltage sensitive nor are there dead spots in the sweep of the pot. If a player just wants a bit of loudness reduction from maximum power, he can have it. If he wants to have better resolution of control at the quiet end, he has it.

With the previous Power Scaling kits one can add a "build out" resistor in series with the PS-0 end of the Power Scale control to eliminate the dead sweep at low loudness. Values up to 33k are useful. Set PS to zero and add 33k in the PS-0 wire to the PCB. Have a listen to how loud this is with your specific amp. If it is too loud, change the 33k to a smaller value, or parallel it with a second resistor to achieve the same effect. In this test, you could use a pot to vary the build-out resistance while playing through the amp, and then measure the portion of the pot used and install a fixed resistor.

Over the recent years, I have been tweaking the values around the pot to try to optimise the sweep. A big problem with that approach is that the mosfet at the input to the circuit, fed from the PS wiper, has variable gate threshold voltage requirements from unit to unit, making the circuit sensitive to B+ (Zh and Zp use a different input device that is far more consistent and predictable). Thanks to Steve Ward, I was reminded of the idea of simply adding a resistor in series with the bottom of the Power Scale control, exactly as was done in Classic-PS and the SB-series kits. It's funny how your thought process can take a direction that slowly limits you, blinding you from seeing that a solution exists right at the beginning of that path. Well, the light came in.

[K O'Connor]

Hi Guys

One of the advantages of cord-wood construction is a reduction in PCB size, which allows kits like SV1 and Sv2 to fit into tinier spaces inside an amp chassis. The disadvantage of cord-wood construction is increased build time.

For example, the SV1-D has a about half of its components lying flat and the rest as cord-wood. All the flat resistors can be installed in one go and then soldered, where each cord-wood mounted resistor has to be inserted and soldered as a step unto itself. The assembly time for the SV1-D is about one hour. It is similar for the SV1-M36, which has a few more parts. SV2-D and SV2-M36 take a bit less time, maybe 45-minutes or so.

The Precision Power Scale SV1-Zh and SV1-Zp take two hours to assemble as there are very few lay-down resistors.

It is a deceptively larger increase in assembly time for the SV2-Zh or SV2-Zp compared to the SV2-D. This is another two hour build BUT once you are done you have both the SV2 and VCK circuits built AND the precision of the Z circuitry. The Zh and Zp represent a complete Power Scale solution for cathode-biased amplifiers and take less board space than the previous combination of kits required.

[K O'Connor]

Hi Guys

I have attached a PDF to show the "recent" evolution of the SV1, SV2 and SV84 kits. There is a size shift both decreasing and then slightly increasing, along with an increase in component density. That density increases the build time and the care required for assembly.

The right-hand family of boards had the ground-referenced Power Scale pot for easy paralleling. The core of the circuit persisted into the middle family of boards, the D-series, and certain issues persisted. The transition saw the number of cord-wood mounted resistors double and board size go down by about one-third in area. The SV84 did not change through this transition.

The SV3-D took the SV3 designation from a previous form that had a quite different function. The original version was a high-voltage expander; a tracking high-voltage circuit that allowed a low-voltage SV to control Vs and bias without directly controlling Va. This allowed all the mosfets in SV3 to be paralleled for greater dissipation for Va control. This form was released and no one bought them. The D-version was designed but never released.

The right-hand family of boards use the Z-core, these are the Precision Power Scale kits. The Zh has no AT trimpot, where the Zp does. SV1-Zh/p is a bit larger than the previous D-series, but it has ideal behaviour, overcoming all of the previous issues. It is as ideal as the SB-series except without the super-expensive pot. The number of cord-wood resistors is up to 27, partly because three single 5% resistors became four 1% resistors each. SV2-Zh/p includes VCK, so it is a complete Power Scale solution for cathode-biased amplifiers. SV84-Zp is smaller than the original SV84 but covers the same class of amp SV84 was designed for. SV7-Zp is the high-voltage big brother to SV84-Zp; both are for single-node output stages and require VCK for a complete solution.