"Sub Zero" Low Voltage Submini Tube Push/Pull Amplifier Project

[Ripthorn]

A lot of the early tweed Fender's did not have a dedicated bias tap and used various methods to get a -ve bias voltage. Might be worth it to check them out.

IIRC they tapped a portion off the HT AC secondaries are dropped the voltage quite substantially with a dropping resistor. Then went on and did the normal bias stuff...

Actually, I think there is something like this that Ken Fischer did on the Express now that I think of it.

[kurtlives]

Ya Ken did the exact same thing. Notice how he drops down that large AC voltage with the large value (220K) dropping resistor.

[Ripthorn]

What kind of current pull is there onto the grid for negative biasing, I wonder?

[frequencycentral]

What kind of current pull is there onto the grid for negative biasing, I wonder?

Not much I would imagine, a few ma at most.

[kurtlives]

Another bump from me...

For the SMPS inductor what was the current rating? Is there any benefit to using a higher value?

Also I would really like to know the current capabilities of the SMPS? I am afraid it wont be able to handle the current draw of a dropping resistor. Need a dropping resistor for my screen source.


For two 5840's what would be a good primary impedance? I am using a Hammond 125B... I am not sure if it is on the date sheet, but I cant find the impedance for one tube, as if it was in a single ended amp.

[Ripthorn]

I noticed the same thing about the 5840 on the sheets I looked at.  If you picked something in the 5-10k range, you would probably be fairly safe.  There's also nothing wrong with getting an inductor with a nice high current rating, I should think.

[Cliff Schecht]

I'll be fascinated to see how your project turns out. And the nature of the 'unique' PS.

IMO submini amps should be submini-small. So SMPS's really go hand in hand with them. I'm really looking for a SMPS that can reliably put out 300 volts before I develop the Sub Zero idea any further, those 7327 tubes are the highest voltage rated I've come across @ 300 volts.

I'd be glad to help you design a SMPS for the general community to use. Since you are talking about higher voltages however, I would prefer to use a current-mode switching IC for the main controller. These devices aren't only inherently much safer than a 555 timer (with peak-by-peak voltage regulation), they have the potential to be very efficient which keeps heat down. The ability to oscillate at fairly high frequencies (I like 100 kHz but you can go higher!) makes for a small PSU (higher frequency allows for smaller transformers). The abundance of reasonably priced parts on Mouser and Digikey makes it almost unexcusable to not use SMPS specific IC's.

What are the voltages/currents that you need available? If you have a bunch of weird voltages that need to be derived, a flyback style SMPS may be your best friend. Getting all of your different positive and negative voltages is as easy as adding the necessary windings to the transformer (really a coupled inductor). Of course, this requires you to wind your own transformer, but I think if we choose a certain torroidal core standard then everyone would be able to wind their own.

[Ripthorn]

Hi Cliff, that sounds like a great idea to me.  I still need to build my SMPS, so I will hold off if this is going to progress.  I would be a fan of something that can put out up 300V or so, personally.  That would mainly be for preamp tube plate voltages for my own use.  I would be interested to hear what Rick and Chris both have to say.

[kurtlives]

Id be into something in that range.

I'd want something that could handle a dropping string with filtering.

With the SMPS Rick posted it can't handle the dropping string I was using. I want a B+ with dedicated supplementary B+s for screen supply, PI supply, and one low one like 50V.

[Ripthorn]

My only concern is that power tubes can have quite a high current draw, and I am not sure what SMPS designs can really do at those high voltages.  If it's possible, I would love something in the 100-200 ma range so that an entire amp can be run off it, but I would settle for enough for the preamp and possibly PI, maybe a screen supply too, but then the question comes up of "what am I going to do with a screen supply if I need a separate HT for the power plate supply?"

[frequencycentral]

I'd be glad to help you design a SMPS for the general community to use.

Cliff, that would be superb. I'm using a 555 based SMPS more through ignorance than choice. Don't get me wrong, they work well, but there must be room for improvement. I'm planning to breadboard that SMPS again and tweak some values to see what happens, but I'm really working in the dark, as I have only the dimmest idea of how it actually works.

What are the voltages/currents that you need available?

I'd agree with Brian, 300 volt, 100 to 200 ma. And the option to include a sub circuit providing maybe up to -50 volts for negative grid biasing. I've got some 5902 beam pentodes which would appear to be the ultimate in subminis, but possibly unsuited to SMPS suppies due to their hunger? Ther 5902WA can handle up to 250v, though the standard 5902 only goes up to 150v.

http://frank.pocnet.net/sheets/127/5/5902WA.pdf

I'd want something that could handle a dropping string with filtering.

With the SMPS Rick posted it can't handle the dropping string I was using. I want a B+ with dedicated supplementary B+s for screen supply, PI supply, and one low one like 50V.

Yeah that too!

The challenge I feel is to decide exactly where is the line at which it becomes unrealistic to drive an amp from a SMPS. For example - is it realistic to ask a SMPS to drive a push pull amp made up of two 5902's in teh power section and two or three 6111/6021/6112 in the preamp? ie how big would the 12 volt supply need to be in order to drive not only the heater requirement but also the current draw of the SMPS? There must be a point at which it's just more practicle to use a power transformer.

But the ultimate purposed designed SMPS for submini tube amps? Count me in! After all, the SMPS I'm using right now was actually designed for nixie clocks........  

[Cliff Schecht]

More on this later, but really as a general rule is if you can power something with a 60 Hz transformer then you can power it with an SMPS, probably with improved results. If I design a supply that does 40 Watts or so then you can use the thing for preamps, power amps, heaters and anything else that needs power in a Sub-Mini design. Again, I think a flyback style design is in order here as it makes it very very easy to add on any positive or negative voltage that will fit on the torroid.

[Ripthorn]

More on this later, but really as a general rule is if you can power something with a 60 Hz transformer then you can power it with an SMPS, probably with improved results. If I design a supply that does 40 Watts or so then you can use the thing for preamps, power amps, heaters and anything else that needs power in a Sub-Mini design. Again, I think a flyback style design is in order here as it makes it very very easy to add on any positive or negative voltage that will fit on the torroid.

Cliff, you are the man.  If I can help in any way, let me know.

[markeebee]

Cliff, you are the man's man.  I probably can't help in any way, except maybe I could hold your coat.

I was thinking about hashing a switcher as well, maybe along the lines of:

http://www.fairchildsemi.com/ds/FS/FSCQ1265RT.pdf

However, I'd be straining at the limits of my capabilities, I think.  And Cliff will probably do it in his tea break.  So I'll stand back, admire, and let somebody else set fire to their shed. 

[MetalGuy]

For the SMPS inductor what was the current rating? Is there any benefit to using a higher value?

The les teh DC resistance, the better which means current rating of ~1A for currents of up to 20-25mA, and 2A for higher currents. Also the less the FET's ON resistance the better. The small ON rsistance TO-220 power FETs however have high prices reaching 15-18$.
Also note that SMSPs introduce high frequency switching noises which later become audible especially if you're dealing with high gain stuff.
A good SMPS would cost more than a regular transformer PSU. A 300V 100-200mA SMSP is serious stuff and will cost even more.

[maarten]

Re 5840: haven't looked at the datasheet, but you should be able to calculate it by dividing the platevoltage by the plate currency; so for an EL 84 you will find 250 volts/0,048 A = 5200 Ohms for the primary winding. In practice you will find that quite some deviation is possible.
Maarten

[Cliff Schecht]

For the SMPS inductor what was the current rating? Is there any benefit to using a higher value?

The les teh DC resistance, the better which means current rating of ~1A for currents of up to 20-25mA, and 2A for higher currents. Also the less the FET's ON resistance the better. The small ON rsistance TO-220 power FETs however have high prices reaching 15-18$.
Also note that SMSPs introduce high frequency switching noises which later become audible especially if you're dealing with high gain stuff.
A good SMPS would cost more than a regular transformer PSU. A 300V 100-200mA SMSP is serious stuff and will cost even more.

Wrong, wrong and more wrong. I'm not even sure what you're talking about in your first statement. The value of the inductance and the power handling requirements all come out very clearly after some back of the napkin calculations. The goal with any switcher is to maintain the V-S balance in the inductor, as the inductance itself is a function of volt-seconds per amp and this must be maintained throughout the switching cycle. If this isn't maintained, the inductor will saturate and your switch will probably burn up.

Power FET's are incredibly cheap now, both in TO-220 form and DPAK's. The on-resistance isn't the only factor that plays in with switchers, you also have the amount of gate charge that is required to turn on the FET. If the gate charge is too high (a severe problem with the lowest on-resistance FET's) then you start getting moments during your on switching where the gate is charging and nothing is happening with the FET. This is wasted energy and is not desirable. Fairchild makes some fantastic FET's that have a nice balance between on-resistance and required gate charge.

High frequency switching noise is NOT audible in any way with a properly designed switcher. You have to choose a frequency much above what the highest audible sound is (about 20 kHz). I like 100 kHz but higher frequency works just as well. If you can hear this, even in your high gain amplifiers, then I commend you for having capabilities beyond that of any normal human. Maybe you should try out for "Heroes"..

I wouldn't recommend a switcher if I didn't believe it was a better option.

[frequencycentral]

Cliff, I've done a bit of research on flybacks, they use a transformer right? How physically large will that component be? I also gather that the transformer may be a sticking point as it's one component that would have to be homemade, possibly using a "certain torroidal core standard" as you suggest. Why does winding my own make me a little nervous? Just fear of the unknown I guess. Is there any way - forgive my ignorance here - that the SMPS could use an off-the-shelf part for the transformer? If so I think it would make the SMPS more universally viable for noobs (like myself) to winding their own.

[MetalGuy]

Wrong, wrong and more wrong. I'm not even sure what you're talking about in your first statement. The value of the inductance and the power handling requirements all come out very clearly after some back of the napkin calculations. The goal with any switcher is to maintain the V-S balance in the inductor, as the inductance itself is a function of volt-seconds per amp and this must be maintained throughout the switching cycle. If this isn't maintained, the inductor will saturate and your switch will probably burn up.

Power FET's are incredibly cheap now, both in TO-220 form and DPAK's. The on-resistance isn't the only factor that plays in with switchers, you also have the amount of gate charge that is required to turn on the FET. If the gate charge is too high (a severe problem with the lowest on-resistance FET's) then you start getting moments during your on switching where the gate is charging and nothing is happening with the FET. This is wasted energy and is not desirable. Fairchild makes some fantastic FET's that have a nice balance between on-resistance and required gate charge.

High frequency switching noise is NOT audible in any way with a properly designed switcher. You have to choose a frequency much above what the highest audible sound is (about 20 kHz). I like 100 kHz but higher frequency works just as well. If you can hear this, even in your high gain amplifiers, then I commend you for having capabilities beyond that of any normal human. Maybe you should try out for "Heroes"..

I wouldn't recommend a switcher if I didn't believe it was a better option.

I'm not very much into theory - what I said was from experience and what other people  designing SMPS said on the subject. I'm more into building circuits, experimenting and listening. Theory is good thing but some soldering is required as well.
The prices of very low resistance TO-220 MOSFETs are not "incredibly low" but I gues it's relative to everyone's budget.
I did experiments with a high gain preamp and several SMPSs comparing them to a regular transformer power supply. Although the high frequency switching noise itself is inaudible The Nixie power supply mentioned above and similar "simple" SMPSs produce a distinctive harshness and "dirt" in the high frequencies which can be distinguished in blind A/B tests without any problems even to the not very experienced year. It's not something drastic but it's there.
The more comlicated the SMSP the less the problem but then arises the question why building a power supply that is bigger and maybe several times more expensive than a regular transformer PS? As you know a good SMSP would require a special transformer. Very few people will start winding their own and the good ones I've seen online are sometimes more expensive than a regular small transformer.
So if you can come up with an easy to build and less expensive than a regular transformer PS SMPS I'll be glad to see that and build it.

[Cliff Schecht]

This is a $0.32 (single quantity) FET made by Fairchild: http://www.fairchildsemi.com/ds/FQ%2FFQP10N20C.pdf - if you can't afford this then I doubt you can even afford to build effects!

As far as that pesky "transformer" goes, it's all about finding an inexpensive but reliable core for us to wind the transformer on. There are different options on the core shape and material, but I really prefer to use torroidal cores. Premade torroidal cores tend to be very expensive because they can't be wound by a machine (not that I know of) but if you wind the part yourself, then really all you're paying for is the core itself and the enamel wire that's needed to wind the transformer. This is where you can start getting saving money and/or getting fancy by using "Litz" wire (smaller strands wound tightly together to counteract the skin effect of any conductor at higher frequencies, google it!) and taking taps off of your transformer for lower voltages, etc.. And these things aren't hard to wind either, in fact it's pretty relaxing to sit down and wind a few up IMHO.

The issue of size vs. cost is a huge factor whenever a manufacturer goes to design an SMPS's board. SMPS's are measured in Watts per cubic inch and as you should be able to see, the larger the number the better. The size and cost of the power supply really comes down to how much power it needs to push and in what form factor. Like I said earlier, it wouldn't be very hard to get a decently small 30 W power supply designed at a reasonable cost. So this leaves us with the question of how much power you guys are asking from the power supply. I'd say that our best bet is aiming at about 30 Watts, this is rated slightly above what most simple two or three tube amplifiers draw (with tube heaters) and would give us a reasonable trade-off between size and cost.

As far as noise is concerned, the output noise is a function of the size and ESR or the output capacitors vs load. Really you see the least amount of noise in a switcher when its running near its full load so our goal would be to choose a wattage that is appropriate for the load. The Nixie tube and similar power supplies aren't designed to be low-noise, they're designed to be cheap power supplies that can power Nixie tubes. The supply that we would have now would be a unit that is specifically designed to be as quiet as possible. Think One-Spot (shameless plug for R.G.) for tubes - you absolutely will not be able to hear the power supply at all.

How much are you currently paying for a 30 W transformer? I paid $30 this summer for a cool old unit that I'm using in a single-ended amplifier I just finished designing and building. I think I can beat that with a switcher... I'm pretty busy this week but my father, a designer of SMPS's for about 30 years, has also taken a vested interest in this project. It may take me a while because we're both busy right now, but the last switcher he helped me on (a unit for PAiA actually) has already proven to be a great benchtop power supply and is quieter than any 60 Hz transformer will be.