Hi,
After nearly two decades my 808 SET is finally operational, with an experimental PSU:
Due to the unusually high Z plate load (5k), I'd like to keep the B+ as close to the 450V rating of the caps as possible but without severe voltare overshoot during startup.
I'm currently using a 310V toroid, 1300V SS diodes and a simpe CLC filter which works quite well but my gut feeling tells me there's more output power to be harvested if the B+ is increased.
The best idea I can come up with using parts I already own would be a 360-0-360V AnTek toroid followed by a 430-440V Zener string and a big Mosfet as a pass device.
IDH tube rectifiers would solve the startup problem but the current draw is close to 300mA (almost 1/3 of that through the cathode follower driver stages).
Any advice for a current production Mosfet (or IGBT, or whatever) that would survive this situation in the long run?
After nearly two decades my 808 SET is finally operational, with an experimental PSU:
Due to the unusually high Z plate load (5k), I'd like to keep the B+ as close to the 450V rating of the caps as possible but without severe voltare overshoot during startup.
I'm currently using a 310V toroid, 1300V SS diodes and a simpe CLC filter which works quite well but my gut feeling tells me there's more output power to be harvested if the B+ is increased.
The best idea I can come up with using parts I already own would be a 360-0-360V AnTek toroid followed by a 430-440V Zener string and a big Mosfet as a pass device.
IDH tube rectifiers would solve the startup problem but the current draw is close to 300mA (almost 1/3 of that through the cathode follower driver stages).
Any advice for a current production Mosfet (or IGBT, or whatever) that would survive this situation in the long run?
This is the beefyest 'superfet' for linear applications. 1A/700V, 0.65A at 1kV SOA!
The input capacity is on the higher side but whatever. It's pricy.
https://www.mouser.com/datasheet/2/308/1/FCH060N80_F155_D-2311638.pdf
The input capacity is on the higher side but whatever. It's pricy.
https://www.mouser.com/datasheet/2/308/1/FCH060N80_F155_D-2311638.pdf
Thanks!
Looks like a beast, though a bit pricey and out of stock at my component dealer.
I could change to 500V caps of course, but I guess my 360V transformer will produce over 500Vdc when the tubes are cold.
Delaying the B+ with a TV damper diode would also be an option.
I came to think of one more thing: My OPTs are 5k/8R but my main speakers are on the low side of 8R, so optimizing the voltage/current for a 5k load might be unnecessary.
I'm planning to build a better prototype PSU this weekend with LCLC filament supplies instead of SMPS, when I'm done with that I'll take a closer look at the B+ supply.
I mean the caps inside the amp. The PSU will be in a separate chassis and any cap between the rectifier and the mosfet will be rated at least 550V (2x385V or so in series).
I don't think I'll need more than 450V into the amp to get a good operating point for the tubes, but the voltage overshoot during startup must be taken care of. Changing the caps inside the amp chassis to 500V ones would be doable but a bit awkvard, and finding 500V caps with the right values and physical sizes wouldn't be easy.
A zener string and a mosfet that clamps the voltage to 430-440V would probably be a decent solution, as long as the Fet doesn't fail and shorts out drain to source.
I don't think I'll need more than 450V into the amp to get a good operating point for the tubes, but the voltage overshoot during startup must be taken care of. Changing the caps inside the amp chassis to 500V ones would be doable but a bit awkvard, and finding 500V caps with the right values and physical sizes wouldn't be easy.
A zener string and a mosfet that clamps the voltage to 430-440V would probably be a decent solution, as long as the Fet doesn't fail and shorts out drain to source.
Ok, at lower anode voltage, and therefore higher gridcurrent, more h2 could sound nicer, i cannot say, i never tried a high amplification transmitting triode that way.
I can understand your worries during start up if the 808 is not preheated before applying anode voltage, voltage limited caps will under this circumstances defenatly need some overvoltage protection.
But would it not be easier, also regarding the 808, to delay the andodevoltage in respect to heater, than adding heatsink, zeners, fet a.s.o ?
Cool locking tube b. t. w.
I can understand your worries during start up if the 808 is not preheated before applying anode voltage, voltage limited caps will under this circumstances defenatly need some overvoltage protection.
But would it not be easier, also regarding the 808, to delay the andodevoltage in respect to heater, than adding heatsink, zeners, fet a.s.o ?
Cool locking tube b. t. w.
Last edited:
The sound quality of this amp is still pretty much an open question, I've only listened to it for a few hours and it soes sound surprisingly good even with zero NFB, which I didn't expect considering the pentode-like plate curves. A bit of NFB will be needed to tame the output impedance, and I'm going to try CFB from the 16R taps first. This should provide 4-5dBs of feedback, which may or may not be enough.
The grid current is provided by a pair of 807s working as pentode-wired, direct coupled cathode followers so there is some soft start built in for the 808s as the grids rise from 0 to ~30Vdc after 30 seconds or so. So far so good, the problem is the voltage rating of the caps.
Any solution that prevents the voltage from overshooting during warmup would work, but the semi-regulated voltage from a zener string-mosfet combo would be benificial to keep the bias stable when the mains voltage changes.
The grid current is provided by a pair of 807s working as pentode-wired, direct coupled cathode followers so there is some soft start built in for the 808s as the grids rise from 0 to ~30Vdc after 30 seconds or so. So far so good, the problem is the voltage rating of the caps.
Any solution that prevents the voltage from overshooting during warmup would work, but the semi-regulated voltage from a zener string-mosfet combo would be benificial to keep the bias stable when the mains voltage changes.
The solution might be much simpler than I first imagined: I just dug through my box of damped diodes and found six or seven quads of 6AU4GTAs, not counting the quartet already employed by my 6S4S PP amp.
My 2x360V Antek toroid has dual 6,3V 5A heater windings that can heat a quad of dampers easily and they warm up slow enough to avoid voltage overshoot during startup. With a decent choke in place and a resistor in series with the cathode on each damper for proper current sharing, I don't expect any problems to tune in the right voltage by adjusting the first cap after the rectifier.
I've seen indications that a single pair of 6AU4GTAs would be able to handle 300mA DC but a quad should have an almost unlimited life expectancy.
I somehow trust tubes more than SS when the voltage goes north of 300V or so, but I think I'll change the caps inside the amp to 500V ones later on just for good measure.
My 2x360V Antek toroid has dual 6,3V 5A heater windings that can heat a quad of dampers easily and they warm up slow enough to avoid voltage overshoot during startup. With a decent choke in place and a resistor in series with the cathode on each damper for proper current sharing, I don't expect any problems to tune in the right voltage by adjusting the first cap after the rectifier.
I've seen indications that a single pair of 6AU4GTAs would be able to handle 300mA DC but a quad should have an almost unlimited life expectancy.
I somehow trust tubes more than SS when the voltage goes north of 300V or so, but I think I'll change the caps inside the amp to 500V ones later on just for good measure.
Hi Fuling
If i understand correctly, you are worried about the 500V limit, set by the caps in the amp. It is a problem because one 808 (even when fully heated before applying anode voltage) will only draw 20mA or so until the kf is heated up and can supply enough positive grid voltage.
As one 808, with +30V on the grid, cannot draw anything near 300mA i guess this 300mA is the total anodecurrent drawn by all tubes from both channels.
So, with preheated 808, this tubes should be able to load the ps instantly with 40mA or so.
How about a simple solution to get some slowly ramping up voltage with help of one, or more, serial connected 300mA NTC between the PS and your amp?
In case you are unfamiliar with those, the 300mA types where used in serial with the 300mA heaters of the P-type tubes in colour tvs.
Offcourse, to be safe and get a clear picture of how it will work out, you should try it out with your lower voltage transformer first.
I can measure the actual cold resistance of those NTCs if you want (I should have one, somewhere...).
Fully heated up the voltage loss should be about the same as a PY500 booster diode at 300mA.
Just one more idea, so you dont run out of options
If i understand correctly, you are worried about the 500V limit, set by the caps in the amp. It is a problem because one 808 (even when fully heated before applying anode voltage) will only draw 20mA or so until the kf is heated up and can supply enough positive grid voltage.
As one 808, with +30V on the grid, cannot draw anything near 300mA i guess this 300mA is the total anodecurrent drawn by all tubes from both channels.
So, with preheated 808, this tubes should be able to load the ps instantly with 40mA or so.
How about a simple solution to get some slowly ramping up voltage with help of one, or more, serial connected 300mA NTC between the PS and your amp?
In case you are unfamiliar with those, the 300mA types where used in serial with the 300mA heaters of the P-type tubes in colour tvs.
Offcourse, to be safe and get a clear picture of how it will work out, you should try it out with your lower voltage transformer first.
I can measure the actual cold resistance of those NTCs if you want (I should have one, somewhere...).
Fully heated up the voltage loss should be about the same as a PY500 booster diode at 300mA.
Just one more idea, so you dont run out of options
Last edited:
You've done your homework, gorgon53
The output tubes draws 20-25mA when the CF drivers are still cold, then increases to 125mA when fully operational. For safety reasons I measure the cathode current rather than the plate current, the real plate current is probably around 105-110mA.
An NTC could be an option, yes. I've never use them in series with the HT but it would be interesting to try some time. For this project I think I'll just go with 4* 6AU4GTAs as rectifiers as mentioned above, since I had several more than I remembered and I've always gotten good results when using them as rectifiers.
I have some RK60 (1641) rectifiers that I bought years ago to use together with the 808s, similar to 5U4G but with dual top caps. Probaby the worst kind of rectifier to use where high surge voltages during startup is not allowed but they might come to good use later on, as I'm already collecting tubes for my next high mu DHT SET
The output tubes draws 20-25mA when the CF drivers are still cold, then increases to 125mA when fully operational. For safety reasons I measure the cathode current rather than the plate current, the real plate current is probably around 105-110mA.
An NTC could be an option, yes. I've never use them in series with the HT but it would be interesting to try some time. For this project I think I'll just go with 4* 6AU4GTAs as rectifiers as mentioned above, since I had several more than I remembered and I've always gotten good results when using them as rectifiers.
I have some RK60 (1641) rectifiers that I bought years ago to use together with the 808s, similar to 5U4G but with dual top caps. Probaby the worst kind of rectifier to use where high surge voltages during startup is not allowed but they might come to good use later on, as I'm already collecting tubes for my next high mu DHT SET
Last edited:
Yes, I used SMPS for the initial tests. They are now replaced with LCLC-filtered supplies using 3x 1,1mH 10A iron core chokes and 5x 22000uF per channel. Still on a breadboard while I'm working on a chassis that has room for those gargantuan filament supplies plus a better HV supply (tube rectified).