• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

New CCS/ depletion mode MOSFETs available... experience anyone?

Status
Not open for further replies.
Jump to Latest
pauldune

pauldune

Member
Joined 2009
#1
Recently, there are a lot more high voltage depletion mode mosfets available then previous.
Like the IXTH6N100D2.
http://ixdev.ixys.com/DataSheet/DS100183A(IXTA-TP-TH6N100D2).pdf
Seems too good to be true. I remember the people trying to get their hands on a 10M45S or a dn2540.
The IXTH6N100D2 is easy to find and dwarfs the old parts specs; 1000v 6A!
Am I missing something?

Has anyone used this (or similar) part as a CCS in tube amps?
Or as a pass device in cascade CCS?

Paul
 
Last edited:
#4
just check for distributor stock

for V regulator use the parasitic C isn't too big a limitation

but for CCS load swinging with full V audio you do want the lowest Id rated part that fits the application to keep the parasitic C down

using a 6A rated part for 30-60 mA just isn't a good match when 100-800 mA rated parts are available too
 
Last edited:
#6
DN2540 is 450V, and low capacitance. There is a TO220 part good for a range of at least 2- 50mA in my experience .The BSP135 is a SMT part also with low capacitance, and goes to 600V but is only rated at 20mA and much less dissipation than the DN2540. The BSP135 is however good down to 100s of microamps, and works well in a cascode.

DN2540 + BSP135 cascode works very well indeed . So does the big Ixys 1kv part under discussion and the DN2540. I have tested this combination; with 50V across the cascode the impedance is still more than 10M ohm at 50khz, so more than good enough for our purposes!
 
#11
Xpersephone2 said:
DN2540 is 450V, and low capacitance. There is a TO220 part good for a range of at least 2- 50mA in my experience .The BSP135 is a SMT part also with low capacitance, and goes to 600V but is only rated at 20mA and much less dissipation than the DN2540. The BSP135 is however good down to 100s of microamps, and works well in a cascode.

DN2540 + BSP135 cascode works very well indeed . So does the big Ixys 1kv part under discussion and the DN2540. I have tested this combination; with 50V across the cascode the impedance is still more than 10M ohm at 50khz, so more than good enough for our purposes!
Click to expand...

This was the information i was looking for; I presume 10m45s in cascode with the big ixys is also good enough?
I need a 1000v 200mA CCS for a parafeed output stage with GM 70.
260watts dissipation per channel for the best 10watt rms imaginable ;-)🙄


Paul
 
#12
if you cascode then both devices don't need to be high V, could even use a settable ccs IC

but cascode doesn't help with too big pass device parasitic C - you're still charging/discharging it out of your output I budget

another option is "srpp derived" modulating cs with the depletion mode mosfet - may need 2 resistors instead of relying on 1/gm
could cut your power in half for the same output
 
#15
I have had good results paralleling the upper device(s) in cascode with a suitable CCS underneath controlling the current. This enables using the 10M45 or IXYS device for more overall current. Of course, then you end up with additional capacitance, so ideal design would determine when to migrate to a different upper device.

I think the important reminder is to not always jump to the highest voltage/current device you can when you want to build something. Pick the device that best fits the application. That's why we don't use 2N3055 when a 2N2222A will suffice.
 
#16
I have found that these devices need to be considered in light of turn on transients which can spike high and take them out of their SOP. This is especially true of direct coupled applications which can behave very strangely on power up/down in ways that valves would just suck up.
I like to give them very safe margins for this reason.

Shoog
 
#19
At these power levels you are going to need to examine the MOSFET SOA carefully. I cannot see any reason parallel cascodes would not work. There will need to be a fan cooled heatsink in the mix, or a huge passive heatsink - look at what trannie Class A amps get built like. The parallel CCS would parallel the impedances. If the top device in the cascode passes AC because of shunt capacitance it gets stopped by the bottom device. Like I said earlier a Ixys / DN2540 cascode still has 10 M Ohm impedance at 50khz.

So my baseline design concept would be 5 off Ixys / DN2540 cascode on a Fischer LAM heatsink, a very big one to keep the noise down.

This will be expensive, and involve a bunch of dead MOSFETs in getting it right. Wear safety glasses, when they explode it's like a fire cracker! 😀
 
#20
disco said:
Why? They have a bank account in Germany if you don't use credit card.
Click to expand...

I don't like to order outside of the EU because of the import duties, taxes and handling cost of the local post service. However, since mouser has an address in my country, you don't suffer any of these. You order locally even though it is send from the US.
This is where I got 10M45 and LM22679 chips which are not available elsewhere here.
 
Status
Not open for further replies.
Top Bottom