I am developing my power supply and did some searching and found 2 ways (are there more?) types of hybrid SS combined with tube rectifier. I am going to use a currently made Russian 5ar4 and i think it is prudent to back it up with a SS helper. The two schemes i have found is either the SSR in series with the high tension secondary or the SSR with the anodes grounded at the high tension input to the anodes of the tube rectifier. If there are other schemes that should be considered please let me know.
If one is superior to the other please let me know and why. Personally i prefer the SSR grounded as that is slightly easier to implement. Also does it make any difference to the sound quality of the PS if the SSR is an el cheapo 1n4007 or Cree Schottky 1200v ? cheers.
If one is superior to the other please let me know and why. Personally i prefer the SSR grounded as that is slightly easier to implement. Also does it make any difference to the sound quality of the PS if the SSR is an el cheapo 1n4007 or Cree Schottky 1200v ? cheers.
The series SS diode tweak provides some added PIV headroom to Sovtek 5AR4s. PIV headroom is the weak spot of the otherwise highly satisfactory New Sensor product. The graphic shows 1N4007s, but I prefer UF4007s. The vacuum rectifier keeps SS diode switching noise out of the B+ rail, but I worry about something sneaking into the power trafo. If necessary, UF4007s can be snubbed into near silence.
Attachments
What's the point of this strategy - what is to be gained?
It would seem that you'll get all the noise of cheap SS diodes + the purveyance characteristic of the vacuum diode. If you can accept solid-state, why not go with something that works and really sounds great - like HexFRED diodes?
Take a look at the IR HFA08TB60 - 8A / 600V diode. There are also 1200V versions, though the 600V should be OK in most all of our work. One downside though, is the failure mode of the HF rect - it will short-out hard if you short the output to gnd.. pretty much converts to a piece of wire which then a-splodes your filter caps unless the fuse is fast enough to prevent it. The peak (repetitive) current is 24A, with a single pulse of 60A allowable.. so you really do have to eff it up to make it short out.
Other than that caveat, they are the +ideal+ sub for a vacuum diode in audio work.
It would seem that you'll get all the noise of cheap SS diodes + the purveyance characteristic of the vacuum diode. If you can accept solid-state, why not go with something that works and really sounds great - like HexFRED diodes?
Take a look at the IR HFA08TB60 - 8A / 600V diode. There are also 1200V versions, though the 600V should be OK in most all of our work. One downside though, is the failure mode of the HF rect - it will short-out hard if you short the output to gnd.. pretty much converts to a piece of wire which then a-splodes your filter caps unless the fuse is fast enough to prevent it. The peak (repetitive) current is 24A, with a single pulse of 60A allowable.. so you really do have to eff it up to make it short out.
Other than that caveat, they are the +ideal+ sub for a vacuum diode in audio work.
Last edited:
reverse recovery current spikes can be measured, hard to make generalizations about audibility when the generated EMI has such an indirect path to unknown sensitivity circuits
of course tube audio amplifier circuits in general should be less susceptible to the EMI than than SS
but minimizing the noise generation in 1st place seems prudent with RC snubbers, better recovery behavior diode selection
but what you want for "tube rectifier sound" is a question: just lots more series R than a SS diode?
of course tube audio amplifier circuits in general should be less susceptible to the EMI than than SS
but minimizing the noise generation in 1st place seems prudent with RC snubbers, better recovery behavior diode selection
but what you want for "tube rectifier sound" is a question: just lots more series R than a SS diode?
Last edited:
Right now, the 8A / 600V HexFRED can be had from DigiKey for ~ $1.25/pc. Here's the catalog listing: HFA08TB60
A pair of these SS parts costs 10% that of a new GZ34 + socket. The main downside, is the loss of the GZ34 / 5AR4 delayed warmup characteristic. Other than that, the hexfred is audibly indistinguishable from the best of vacuum diodes.
Thanks Eli, i appreciate the diagram. Is there any reason to use your scheme vs grounding the SS diodes?
In regards to why tube rectifier? I have lots of them so why not use 'em. And in particular the 5ar4 is also needed to achieve the B+ voltage i am looking for with my SE amp. Slow voltage turn on/ ramp up with indirectly heated types. I measure 20 + seconds for Mullard 5ar4, which means that the capacitor voltage rating does not have to be overrated to allow for the initial voltage spike until the power tubes kick into gear.
So, i am just wondering about a hybrid tube rectifier scheme.
In regards to why tube rectifier? I have lots of them so why not use 'em. And in particular the 5ar4 is also needed to achieve the B+ voltage i am looking for with my SE amp. Slow voltage turn on/ ramp up with indirectly heated types. I measure 20 + seconds for Mullard 5ar4, which means that the capacitor voltage rating does not have to be overrated to allow for the initial voltage spike until the power tubes kick into gear.
So, i am just wondering about a hybrid tube rectifier scheme.
The point is that existing units employing 5AR4 rectifiers can be fitted with affordable current production tubes, without experiencing "fireworks". OS 5AR4s are obscenely expensive.
The vacuum rectifier blocks SS diode switching noise from entering the B+ rail. The SS diodes provide exactly what is needed, extra PIV headroom. The whole is greater than the sum of its parts.
I will not suggest vacuum rectification of B+ in new power amp builds. High PIV Schottky diodes are every bit as quiet as vacuum rectifiers and they (FWIW) get my nod. While FREDs are somewhat less noisy than UFnnnn parts, they can be problematic to snub. OTOH, people have had great success in snubbing UFnnnn diodes.
If you use a hybrid bridge rectifier to process the end to end voltage of the power trafo, while not connecting the CT, the B+ rail voltage will be about 2X of what is currently present. However, the available current will be halved. TANSTAAFL!
For a vacuum rectifier, the forward drop in a 5AR4 is small. To obtain a "taller" B+ rail and retain current capability, replace the 5AR4 with 2X 1200 PIV Schottky diodes.
For a vacuum rectifier, the forward drop in a 5AR4 is small. To obtain a "taller" B+ rail and retain current capability, replace the 5AR4 with 2X 1200 PIV Schottky diodes.
Last edited:
TANSTAAFL, that doesnt sound good.
I actually did put 2 schottky 1200v in an octal socket and plugged it in and it poofed the 3A fuse.
I would prefer to keep the 5ar4 for the slow start up rather than using a CL.
Have you tried using the SSR to ground method? I did and the voltage were really high llike more than 500vdc. And the CT of the power trans is supposed to be disconnected right?
If i make my filter supply choke loaded with the SSR grounded, (that would make the filter 6H x 50 dcr , 100uf, 6H x 60dcr x 100uf) with the power trans outputting 300-0-300. What would the expected B+ be at C2? I tried modelling this on PSUD2 but the voltages shown were really different from what i measured so i don't know if i am inputting all the parameters.
I actually did put 2 schottky 1200v in an octal socket and plugged it in and it poofed the 3A fuse.
I would prefer to keep the 5ar4 for the slow start up rather than using a CL.
Have you tried using the SSR to ground method? I did and the voltage were really high llike more than 500vdc. And the CT of the power trans is supposed to be disconnected right?
If i make my filter supply choke loaded with the SSR grounded, (that would make the filter 6H x 50 dcr , 100uf, 6H x 60dcr x 100uf) with the power trans outputting 300-0-300. What would the expected B+ be at C2? I tried modelling this on PSUD2 but the voltages shown were really different from what i measured so i don't know if i am inputting all the parameters.
I would then need to install an extra socket and another filament trans. That is why i am trying to stay with what i have.
I just wanted to add that the most recent Russian made tube rectifiers are FAR better than the earlier ones. A Genalex reissue U77/GZ34 or a Tung-Sol reissue 5AR4 are both quite good and far more durable than the Sovtek or Chinese tubes - especially where the rectifier has to work hard. I'm not saying the Sovtek/Shuguangs aren't okay in some cases, but without the diode mod Eli posted they just aren't rugged enough for many applications. I've found the Genalex and Tung-Sol reissues are much stouter - and they are FAR less costly than the increasingly rare old stock.
Last edited:
Combine that with the issue of filament supply and you have something that's (IMO) unsatisfactory. The behavior of hybrid bridge rectifiers is dominated by the "hollow state" stuff.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.