Replacing 5U4GB with solid state recitfier in a SF Deluxe Reverb amp - any mod needed?

[Pixel]

I had a brief encounter with microwave oven diodes when replacing a selenium rectifier. I thought I might avoid needing a dropping resistor if I used the microwave oven diode. It turned out that the diode performed better than expected and had a much lower voltage drop than the graph on the datasheet indicated. So I just used a 1N4007.

[slidincharlie (Carlo P)]

Mars Amp Repair thanks for the suggestion buddy, but the whole thing goes far beyond my very limited tech capacities. I am only able to copy procedures that are explained or written step-by-step.
Someone else here may well give you some feedback about your proposed approach.

[Chuck H]

Does this not seem like a viable solution worth trying?:
I have an idea for replacing the tube rect with SS diodes with the multi-diodes from a microwave oven, which have a substantial drop.;

I just looked into these because there are no specifics for components data in your post. These seem to be multiple diodes in a single diode case making for multiple standard diode voltage drops (.7V). They do look to be affordable enough but the voltage drop looks to be only about 3V for each unit. To make a meaningful difference it would take several of them and the connections might make implementing that a little clumsy?

I was equally vague above when I mentioned a "string of zeners". So I'll try to clarify. A string of 10V 5W Zeners in series with the center tap of the HV winding (assuming full wave rectification) will drop about 8V per diode. This according to Rob Robinette's site and I think I noticed this also when I tried it. You can get three Zeners on one of those five terminal strips that I like to use sometimes. I think we've all seen them. The ones with the the middle terminal that has a mounting bracket. So skip that terminal. A string of three Zeners would drop 24V. If you need more than that you can use two terminal strips face to face on the same screw mount for seven Zeners staggered across the available terminals. At which time you can choose the voltage drop on the diodes for whatever you need. 10V zeners would drop 56V.

As I recall this thing gets hot so don't mount it near electrolytic capacitors and leave the leads long on the Zeners, elevated off the terminals so they don't unsolder themselves.

[Iplayloud]

Does a DR really sag though? I think the 6v6 pair would barely tickle the very efficient 5U4.

A GZ34 would possibly just what you need. The extra 15v would give you bit wider freq response and stay cleaner louder.

Multipliers:
NOS 5Y3 1.1
New 5Y3 1.25 (5v4)
5U4 1.25
5AR4 1.3 (GZ34)
Diode 1.414

[Greg Robinson]

As Chuck explains, microwave diodes aren't a great option for dropping voltage. They're just multiple series silicon diodes in series in a single package, they only drop several volts, and that only at their rated current. If you want stiff voltage drop, power zeners or better, amplified zeners are the way forward. Power zeners are expensive and a bit of a pain to heatsink, but mosfets are cheap and easy to heatsink. Check out the last paragraph of RG Keen's Mosfet Follies. Note, RG failed to mention in the article, when selecting a mosfet, check the datasheet SOA for DC operation. If the datasheet doesn't have it, the fet is intended for switching applications and almost certainly inappropriate for the task. You may have to search through a fair few datasheets as linear rated fets are less and less common these days, but they do exist.

[loudthud]

Does a DR really sag though? I think the 6v6 pair would barely tickle the very efficient 5U4.

This is a simulation of a typical class AB guitar amp. The two numbers next to current loads I1 and I2 represent plate and screen nodes at idle and full load. You can see the sag in the screen supply is mostly the result of increased ripple on the main B+. The choke filters out most of the ripple at the screens. If you don't like the numbers I used, run your own simulation or post what you want here and I can run another simulation.

[Iplayloud]

This is a simulation of a typical class AB guitar amp...

Nice !!
Wondering how it would look like with diodes too. And also 6L6s :-)

[Chuck H]

I use PSUDII as well. I didn't run the requested simulations but the diodes will sag less. That said, a lot of the sag is the result of the DCR of the PT. And you know I love you LT but in my humble experience 50 ohms for the PT DCR might be more ideal than reality. I just looked up the DCR for the Hammond DR PT (since Hammond is one of the only winders that publish this spec) and it's 114 ohms. That additional 64 ohms actually makes for a lot more sag.

[Helmholtz]

I just looked up the DCR for the Hammond DR PT (since Hammond is one of the only winders that publish this spec) and it's 114 ohms. That additional 64 ohms actually makes for a lot more sag.

That DCR in the Hammond spec is for the full HV winding.
As only one half is conducting at a time, only half of the resistance matters.
But effective transformer resistance is the sum of secondary and reflected primary resistance (around 28R).
Using the Hammond values I get around 85R. This will somewhat increase when the PT is hot.

For comparison the effective resistance of a 5U4GB is around 190R (GZ34 is only 68R).

To simulate the sag at full (clean) output one needs to know the full power DC current draw.
I estimate this to around 120mA.

[Chuck H]

Thank you Helmholtz. My understanding of things improves with every post you make. I can see now that LT already knows what you just explained since 50R would be a better general value for the simulation. Unless...

Look at the diagram above for PSUDII. It shows both halves of the winding and indicates the DCR. I have to wonder if the CAD simulation program takes into account that only half that secondary DCR is applicable.?. In other words, should I be plugging in half the DCR for my own simulations? How accurate is the simulator to the values shown? etc. Having used PSUDII for a couple of builds I've actually found it to be accurate with actual bench measurements. But I'm going to do some math now to check this.

[Helmholtz]

Look at the diagram above for PSUDII. It shows both halves of the winding and indicates the DCR. I have to wonder if the CAD simulation program takes into account that only half that secondary DCR is applicable.?. In other words, should I be plugging in half the DCR for my own simulations?

I assume PSUD2 (AFAIK not based on LT Spice) expects the user to enter the effective resistance of the active winding, i.e. the half-secondary.
But I didn't verify.

[Chuck H]

I assume PSUD2 (AFAIK not based on LT Spice) expects the user to enter the effective resistance of the active winding, i.e. the half-secondary.
But I didn't verify.

I just verified. It's as you stated. The program expects the user to indicate the effective resistance. So for a full wave rectifier you do need to figure the primary DCR in series with half the total secondary windings DCR.

That's pretty clunky for a CAD program I think. Since it allows users to switch between rectifier types on the fly I would think the programmers would take something like this into account. Good to know this now though.

[slidincharlie (Carlo P)]

Going BACK to my original question:
I have turned the amp on with the s.s. rectifier and no power tubes. The B+ is 423VDC. Shouldn't the 500 V-rated filter caps and the JJ 6V6 tubes be in safe operating conditions?

[loudthud]

I think there is a forum on Duncan's website where you can ask questions about the operation of PSUD. IIRC I ask the question about resistance spec of a center tapped winding and got the answer that you use half the DC resistance... but there is also a way to take into account the primary resistance and the turns ratio.

I must admit I just guessed at most of the numbers used in the simulation. My main point was to show how the change in ripple Voltage is translated into sag at the screen node by the choke. If you take a working amp and use it's transformer, choke, resistor and cap values and current numbers, you'll see PSUD-II gives fairly accurate results.

[Helmholtz]

. but there is also a way to take into account the primary resistance and the turns ratio.
.

The formula is shown when you click the help button in the transformer properties window and then open the source impedance calculator.
If e.g. the unloaded half-secondary voltage is 360V with a primary voltage of 120V, the measured primary DCR has to be multiplied by (360/120)² = 9 to get the reflected primary resistance (follows from transformer theory).
A 230V primary will have a higher DCR but a lower step-up ratio. I'd expect the reflected resistance at the secondary to about the same as with a corresponding 120V PT.