MosFET or other semiconductor substitute for cathodyne?

[mondogenerator]

The largest DC link stuff I typically see is 8 to 30 uF at 4kV lol, and these are the size of a drinks can.

I have some 15uF PP films which are half the size, and if I recall 1500V rated, some of the old RS (maybe Maplin) PP range.

I have a Monacor branded 250V rated 47uF PP film that might be worth a try, if they're still available.

I can't find them, but I found this:

Not cheap...

C4AQLBW6130A3NK Capacitor Polypropylene DC-Link 130uF Leads for KEMET: Amazon.co.uk: Business, Industry & Science

 

[andyjevans]

The ones I used are sold by RS. They're all box shape so need a bit of thought about mounting them. I use M3 hexagonal nylon standoffs with solder tags on top.

https://uk.rs-online.com/web/p/polypropylene-film-capacitors/6842295/

Ale does some PCBs on Bartola Valves for the WIMA ones

WIMA DC Link – Bartola(R) Valves

I'm very pleased at the improvement in sound I'm getting.

 

[egellings]

How about source-odyne?

 

[kodabmx]

Off topic, but since we're on the subject of bypass caps here's a brief shootout I just did on the cathodes of my 300b outputs. In order of preference:

- Vishay DC Link Cap, 40uF, 900v
- ICW SA Clarity Cap, polypropylene, 47uF, 630v
- RS electrolytic 100uF, 450v
- ICW PW polypropylene cap, 40uF, 160v
- Aerovox motor run cap, polypropylene, 40uF, 450v

My current choice of cathode bypass (only one in the amplifier, the first cathode)
Yunxing SPT1CM102G12OR
https://datasheet.lcsc.com/szlcsc/2007091907_Yunxing-SPT1CM102G12OR_C698129.pdf
I am waiting for a 1500uF version from a different brand to test it against but I don't think it will be necessary at all.

FYI: I also tried Nichicon UKA1C102MPD but there was no noticeable difference.

 

[stocktrader200]

you can couple the fet concertina with .22uf cap and use 3.3M B+ and 1M Gnd resistors to bias it, removing the need to optimize the plate voltage for the concertina

 

[andyjevans]

Hi guys. I now have to build a push-pull amp for a friend who's waiting for it, so I have to have a working design for a phase splitter ASAP. Being only used to tubes I don't fully understand how to design this mosfet splitter and need your help so I can get building. I have some IRFBC20 which Eli Duttman recommends. I now have to complete the following design. I need to know the voltage going into the gate for a start. Then the values for the 2 resistors top and bottom of the mosfet. I've put in 5K but suggest something better. And anything else needed in the design? What diodes should I use? Output stage is EL12n in push-pull, easy to drive.

Thanks for your help in finishing off this design!

 

[andyjevans]

Would this work?

 

[tikiroo]

Yes. Voltage at source will be about 111V, so current through mosfet and resistors ~11 mA, mosfet dissipation ~1.1 watts. Reverse transfer capacitance is at a minimum at Vds>20V, which will be the case for all operating conditions. You probably don't need that much current to drive EL12N grids (pentode connected?), so resistors could be say 22k for 5mA, but up to you.

 

[pblix]

Here is a question more than a comment: Are the Zener diodes needed if the load is not inductive? I know they won't hurt, but are they necessary?

 

[Merlinb]

"Here is a question more than a comment: Are the Zener diodes needed if the load is not inductive? I know they won't hurt, but are they necessary?"

Yes, it is possible at power up for the gate to be pulled high while the source is pinned low as it charges the coupling cap through the grid current of the output valve, busting the gate open. (You really only need one Zener, though back to back Zeners don't hurt of course)

 

[andyjevans]

Yes. Voltage at source will be about 111V, so current through mosfet and resistors ~11 mA, mosfet dissipation ~1.1 watts. Reverse transfer capacitance is at a minimum at Vds>20V, which will be the case for all operating conditions. You probably don't need that much current to drive EL12N grids (pentode connected?), so resistors could be say 22k for 5mA, but up to you.

EL12n are triode connected. Thanks - I might go with 15K resistors, have some nice wirewounds.

If I'm using IRFBC20 do I need heat sinks for them? It would be around 110V at 7.3 mA.

 

[tikiroo]

IRFBC20 junction to ambient is 62 degrees per watt. So it will get hot (too hot to touch) but would likely be fine unless ambient is really high, maximum junction temperature is 150 degrees C. A small clip-on heatsink wouldn't hurt.

 

[andyjevans]

I'm coming to the end of my PP EL84 build on a Leak Stereo 20 chassis. Here is the proposed circuit. The EL84 output stage is up and working. Now need to do the 37/76 input tube and the mosfet splitter. I have some AOT1N60 I propose using.

Any comments on the circuit welcome!

 

[pblix]

Enhancement mode MOSFET, if I'm not mistaken need a bias network on the gate (voltage divider). Look at Tubelab's 100 buck challenge guitar amp schematic for ideas how it can be done.

 

[wg_ski]

With that much voltage on the cathode resistor, it doesn’t really make a difference if the FET is E-mode or D-mode. Just has to handle the voltage and have its capacitances remain linear. Just keeping VDS up near 50V or more will do this with just about any HV FET. Some do better at lower VDS so depening on type it could go lower transparently.

 

[jjasniew]

Ok, what would happen if you grafted the mosfet right across an existing tube; would it simply "override" the tubes operation by brute force? Like the triode sayin; "My friend here says we will drive 10k plate and cathode loads".

 

[wg_ski]

Well, in order for an E-mode to start conducting, vgs would be >0. That would be driving that triode into grid conduction so it’s not the most linear operation. With D-mode you’d probably have either the FET dominating, or the tube, depending on Vth and the resulting Ik.

 

[jjasniew]

OK, sounds like not such a good idea...

 

[Merlinb]

Enhancement mode MOSFET, if I'm not mistaken need a bias network on the gate (voltage divider).

It's directly coupled, so no additional bias network needed

 

[b_force]

For that reason a NPN and probably PNP would also work just fine.