• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

Mosfets for Tubelab SE?

Status
Not open for further replies.
Jump to Latest
M

mctavish

Member
Joined 2008
#1
Its probably been addressed somewhere but my searches have been unsuccessful... Digikey is out of 2SK3563Q-ND, is there a suitable replacement that anyone can suggest? I am moving overseas in a few weeks and I am trying to put together a BOM for another build before I leave. This will be my second TSE and I am using 45's again, George this is a fine piece of equipment!!
 
#5
I have the following mosfets available:

2SK3563
2SK3742
2SK3564

What would you use it for a 300B with B+ around 390 V ? (Simple mode)
What mosfet characteristics are important for this application ? The maximum ratings are obvious, not the rest (Learning mode)

Additionqlly, I want to have separate PSU for the power drive. What current do I have to make available ? I was thinking of using 150VAC with bridge rectifiere and CRC filter.

Thanks,

Davide
 
#9
You want a low gate capacitance (Ciss) that is also constant over the source-drain voltage range that you are operating in.
Click to expand...

In a typical mosfet amplifier this would be true, but the mosfets in the Tubelab SE are used as source followers. This is about the only application where I would use a mosfet in a tube amp.

In a source follower the gate to source capacitance (Ciss) isn't important unless it is ridiculously high. In a follower the Source follows the Gate signal and are at nearly the same AC potential so the Ciss is effectively bootstrapped out of the picture. The Output Capacitance (Coss) appears directly across the output, so it can be important, but the fet should have no problem driving this capacitance. The Reverse Transfer Capacitance (Crss) is the gate to drain capacitance. In a follower the drain is at AC ground so this capacitance appears directly across the input and must be handled by the driver tube. We want this capacitance to be low and constant over the voltage seen across the fet. The Tubelab SE has several hundred volts across the fet. The Toshiba 2SK3563 has constant 7 pf for any voltage above 20 volts, so this one is the one to use here, and its what's in my board.

The data sheets you included do not have all of the graphs. I have the full data sheets, but they are too big to include here. There should be a graph showing the three capacitances VS voltage. Look for a Crss that is low and flat.
 
#10
Thanks for the explanation, George. Based on what I've read in several of your posts, the 2SK3563 is the ideal device for this purpose.

However, the issue is that the 2SK3563 in no longer available through the larger distributors (Mouser, Digikey, Allied, etc.). I don't think it's a temporary production issue; it appears that it's been discontinued by Toshiba.

I don't believe that anyone is second-guessing your use of the 2SK3563, but rather that we're at a point where we need a reasonable sub, or at least guidelines in choosing one. Your post helps a great deal in that regard.
 
#13
A related question for mostly educational reasons.......In a Tubelab SE (or similar source follower application) if the rectified off load voltage of the transformer is greater than 500V, is that a problem for 500V rated parts such as the 2SK3563?
 
#17
we'll see

I'm in the midst of building a tubelab SE and decided to try the IRF830APBF-ND. I didn't notice the FDPF5N50NZ until after I placed my orders.

It's hard to tell from the data sheets which one might be better. The graph of the gate capacitance vs voltage only goes up to 30 volts on the fairchild data sheet. It looks flat until that point, but what happens after that one can only guess. Can we assume that it is flat past 30 volts ?

The gate capacitance of the Vishay chip is lower at 4.3pf, but the relation to voltage not near as flat as the Toshiba mosfet, nor as flat as the fairchild seems to be. At 20 volts it's at just above 4pf and by the time it reaches 100 volts its a little under 3pf.

In his earlier post, George mentioned that low gate capacitance is a criteria in this application. Is lower better ? And which is more important, a lower capacitance, or a flatter curve ?

What will be the effect of the decline in capacitance as voltage goes up on the overall performance of the amp ?

Is the capacitance decline so large as to make the vishay chip inappropriate to use here ?


http://www.vishay.com/docs/91061/91061.pdf
http://www.fairchildsemi.com/ds/FD/FDPF5N50NZ.pdf
http://www.toshiba.com/taec/components2/Datasheet_Sync//75/10801.pdf
 
#19
no subscription

thanks for your thought, Russ. If I were more familiar with the arguements about phase distortion and capacitance slopes I might subscribe to one. But I'm not so...

I looked a little more closely at the data sheet for the Fairchild mosfet and saw that its typical gate capacitance is 4pf, not 8pf. 8pf is the max.

And the scales are so very different on the two manufacturer's data sheets it is very hard to compare the slopes of each chip. Vishay's logarithmic scales exagerate the slope, and Fairchild's chart only goes out to 30 volts, and its vertical scale is linear so the difference between 4 and 3 would be really hard if not impossible to see. Looking closely at the Fairchild chart there is clearly a slope all the way out to 30 volts.

So it looks to me like neither one is the ideal replacement for the Toshiba, neither one has the same flat slope. The two suggested alternatives look very similar to me, the typical gate capacitance is practically the same. I know the slope of the Vishay is only about 25% over the relevant voltage range and I don't know for sure the slope of the Fairchild, so I am going to stay with the Vishay and cross my fingers.

And I guess endure the slings and arrows of the consequences of a -25% slope in my gate capacitance.

I can always rip them out later when we find a better alternative!
 
#20
Success !

Well I finished the Tubelab SE with the Vishay MOSFETs and so far things seem fine. No fires, no detectable leaks of radiation.

And it sounds wonderful.

I brought the amp to life without a choke or oil cap. I wanted both to hear the difference and to keep things simpler. All the voltage readings made sense; my b+ was on the high side at 320 volts so I biased the 45s at around 25 ma. I ran it this way for around 10 hours.

Then I pulled R4, and for it substituted the Triad 150 ohm choke, I changed C4 to 4.7 uf, and added a 50 uf ASC oil cap in parallel with C5. Everything got better: b+ dropped to 308 volts or so, the tiny bit of hum previously present vanished, and the bass response was vastly improved. Where the bass signal distorted and flabbed out it became complete and full. With the lower b+ I reset the 5842 plate voltage again to 175 and biased the 45s at just under 27 ma.

This amp is now absolutely humless and silent when there is no audio signal and the volume is at full. And it sounds great. I had never listened to a SET amp before let alone a DHT SET. Now I know what the hubub is all about. Everything I listen to sounds new again and I find things I hadn't heard before.

I would like to have another Tubelab SE next to this one so I could comment on the difference in sound, if any between the Vishay MOSFETs and the Toshiba's. I can say that the Vishay MOSFETs do sound good and that I have yet to find any bad side effects.

This amp sounds great and is behaving exactly as I expect it to.
😛
 
Status
Not open for further replies.
Top Bottom