A great replacement diode for guitar amps: the Vishay SF1600–and perhaps some digressive philosophical background discussion…

When it comes to guitar amp electronics, there are two thought paths. On the one hand, the 1960s and 70s are considered to be the golden age of tube guitar amps. The ones that I am aware of that are considered by many to have seemingly magical powers and are collectible are Marshall ‘plexi’s, Hiwatts, and Fenders. The Marshall amp had some nice distortion and was used by almost all of the classic hard rock guitarists–Jimi Hendrix, Jimmy Page (Zeppelin), Ritchie Blackmore (Deep Purple), and Clapton. The Highwatt amps had better build quality but were designed for a powerful clean tone. They would not distort until driven to the highest level, and then you got the sound of Pete Townsend (The Who).

Fender amps had a generally clean tone, and were ubiquitous among American guitarists. Leo Fender sold his company to CBS in 1965, and CBS was considered to have implemented cost saving measures on its products, leading to the term ‘pre-CBS’ and creating a golden age and collectability for the older amps. Hiwatt went out of business shortly after its founder, Dave Reeves, died in 1981. Marshall was not thought to have lowered its quality, but the circuits and parts were changed, and when the faceplate was changed from poly carbonate to aluminum in 1969, the so-called ‘plexis’ (plexiglass) became collectible. I personally liked the sound of the JCM 800 series from the 1980s, but nonetheless the perceived magic was in the plexis. The aura of magical powers surrounding these amps were in no small part due to the great music of that era, but whatever the reasons, there is a golden age associated with many guitar amps produced in the 1960s and to a lesser extent the 70s.

So the two thought paths are that the original parts must be used if at all possible, the other being that some improvements can be made with newer parts that are better performing. I share my opinion here, knowing that others may vociferously object to my thinking–that upgrading parts is good in many circumstances–unless the amp is considered 100% collectible and its operation as a ‘player’ is not important.

Getting back to the diode replacement business… I have a Highwatt 50 watt custom that was built shortly after Dave Reeves died, branded as a ‘Biacrown’ model. I note it was still wired by Harry Joyce, which is a wonderful thing. It did not have the legendary Partridge brand transformers, and the Biacrown era amps are not considered very collectible. So, I set about to upgrade some of its parts. Replacing the electrolytic caps was a no brainer, but the older diode technology has always concerned me.

Of relevance to this discussion–solid state technology has improved a lot since the 1960s, certainly in the realm of diodes. The devices available then did not have the capacity for the 600 volt levels driving the power tubes, so they were built with two or three diodes in series. I like to replace these with newer diodes. The diode in the older tube amp circuits needs to withstand a peak reverse voltage twice the ac peak value from the transformer. So, if the transformer puts out 500 volts rms, the peak voltage is 500 times the square root of two (roughly 1.414) which is 707 volts–doubling to 1,414 volts!

I discovered that finding a suitable replacement diode was not as easy as I thought it might be. There are not a lot of diode models for this high voltage. I found one model but got hung up in a discontinued product/back-order situation. I settled on an excellent device, the Vishay SF1600. It has a peak reverse voltage of 1,600 volts, a peak forward surge current of 30 amps, and a 1 amp average forward current rating. Sufficient for my Hiwatt with plenty of over-specification to help reliability. The diode itself is tiny at 4 Millimeters long, and costs 63 cents apiece if you buy 10 of them from digi key.

It also has a 75 nanosecond reverse recovery time. Reverse recovery time measures a little chirp of conduction the diode makes as it switches from conduction to non-conduction during the 60 cycle ac input. The 1N4007 technology reverse recovery time is 1-20 microseconds. To the extent that this produces an audible difference in a guitar (or even hi fi) amp is debatable, but it’s another attribute that has been improved up by the SF1600.

The very popular 1N4007 runs maybe 17 cents, but for the small amount of work I do, the price difference is trivial and there is no reason to stock two different diodes for tube amp rectification, so the SF1600 is my go-to high voltage diode. I use it in my Hiwatt as well as my capacitor multipliers.

An additional note. When ordering another batch of high voltage diodes, I ordered 30 BY448CT-ND diodes at 58 cents each. The differences I could find from the SF1600 are this doide handles two amps as opposed to one for the SF1600. It also has a slightly lower reverse recovery time, but still in the nanoseconds, so these do the job as well.

I had a PCB custom made for my capacitor multiplier–as they said in Flashdance, What a Feeling!

I had built a capacitor multiplier for use in my Hiwatt Custom 50 amp. The idea was to reduce the 60 cycle hum it had. One of my frustrations with the task was how fussy the cap multiplier was to wire, and it looked uncomfortably homemade. I recently noticed that there is free software on the internet that allows one to design printed circuit boards (PCB) and have them built by prototype companies–I settled on KiCad as the design program.

The minimum PCB order quantity is three at oshpark. I ordered three PCBs from oshpark for $31 dollars, and received them today. I assembled one and and have been testing it on my bench. The first thing I did to verify my design was to assemble it according to the printing I put on it and see if the circuit is correct–it was. I would not have been surprised if something was wrong with my first try, and I was prepared to go through a few cycles of prototyping. A properly working design right out of the gate is on the high end of possible outcomes.

On the test bench I have around 500 volts going into it. I have a 1 watt (470k) resistor for a load and its drawing 1.1 ma. It has about 1.5 volts drop from in to out. Perhaps because the low is so low, its dramatically reducing the characteristic sawtooth wave that emerges from a linear high voltage guitar amp type rectifier and capacitor circuit. I note that sawtooth has some high frequency noise riding on it, not sure where that is coming from.

Since the power supply on my Hiwatt can emit in excess of 600 volts. Any apps using small size capacitors need two caps in series, as 400 volts is the voltage limit on readily available capacitors–500 volts is the general limit, and that does not allow for generous over-engineering to make it more reliable. I get my electronic parts from digi-key and mouser. The active device is a BUL416T NPN transistor which can handle 800 volts across its collector-emitter. These are made for fluorescent light ballasts and cost about $3.60 each.

I want to acknowledge Merlin Blencowe for his wonderful book “Designing Power Supplies for Valve Amplifiers” (2010) for the idea and circuit.

I ordered some high voltage FETs, and I am thinking of laying out a PCB using a mosfet in a capacitor multiplier and see how that works…

This is the device itself, very small–around 2×3 inches. Looking at it, I can see that I could have packed the parts more densely for a smaller device. I hope it will fit in the Hiwatt. I noticed the ones I built by had were smaller. For instance, the resistors and diodes were vertically mounted and the caps were closer together. I could also put the silkscreen message with my name on the other side.
Maybe due to the low load, but the output is surprisingly smoother than the input. The idea of this device is to use a transistor in a circuit which simulates an extremely large filter capacitor–roughly speaking. I see some sort of high frequency noise on that sawtooth waveform, not sure where it is coming from. Note that I had to use a 50 times higher range to show any ripple at all in the output signal. I believe under full load, the output ripple will be maybe 1/10 of the input sawtooth wave.
Here is the test setup. The meter on the far left is measuring the dc voltage coming out of it, 502 volts. The meter on the far right is measuring the current through it, 1.1ma. And the scope on the top is measuring the waveform going in and coming out of it. Smoother is better. The device itself is buried under a bunch of wires in the center of the workbench. There is also a bunch of junk on the workbench not related to this activity.

I am currently working on the layout of a more complicated device–the circuitry for a complete high voltage power supply which I will use to experiment with guitar amp circuits.

There is the schematic of the device as I entered it into the KiCad design tool. The high voltage made it simpler to put two capacitors in series. The resistors by the caps are to ensure equal distribution of voltage among the two despite the inevitable differing leakage currents in the caps. The diodes are there to protect the transistor from power on and power off voltage transients. Other than that, it’s an extremely simple circuit but it does the job surprisingly well.

This is what it looked like in the KiCad PCB layout tool.