tom and others who know about switchers,
I was wondering if I could trouble you for some advice. I am trying to do something similar for a quad of 801As. Heater voltage is 7.5V@5A total so the particular regulators that you chose won't work(due to voltage). Does LM25088 look like a good choice?
http://www.national.com/ds/LM/LM25088.pdf
Besides the voltage and current requirements, I wanted to have a soft start input to let the voltage rise slowly like you did. Unfortunately, I couldn't find a capable regulator with soft start with transistor built in. Oh, well. Maybe you guys know of one.
If it looks like a good one, then I have a few technical questions about how to design the support circuitry. I just need a bit of hand holding the first time I do this.
I was wondering if I could trouble you for some advice. I am trying to do something similar for a quad of 801As. Heater voltage is 7.5V@5A total so the particular regulators that you chose won't work(due to voltage). Does LM25088 look like a good choice?
http://www.national.com/ds/LM/LM25088.pdf
Besides the voltage and current requirements, I wanted to have a soft start input to let the voltage rise slowly like you did. Unfortunately, I couldn't find a capable regulator with soft start with transistor built in. Oh, well. Maybe you guys know of one.
If it looks like a good one, then I have a few technical questions about how to design the support circuitry. I just need a bit of hand holding the first time I do this.
I would use this, an LM338:
http://diyaudioprojects.com/Technic...M338-5-Amp-Adjustable-Regulator-Datasheet.pdf
It will handle 5 amps, but you will need a good heat sink.
You may want to look at the MIC5156, it will drive a logic level FET.
MIC5156/5157/5158 Super LDO™ Regulator Controller
MIC5156/5157/5158 Super LDO™ Regulator Controller
Soft start is a good thing. I set mine for a 5-second start-up time and the regulator doesn't even current limit when turned on into a cold filament of a 300B. I like that.
The LM25088 will probably work. I see no reason why it wouldn't. However, note that it's in a package that'll be a bit tricky to solder. Also note that it uses external switches and inductor.
For 5 A output currents, you probably won't find parts with internal switches. Even at 1 A, they start having at least one external switch (Schottky diode). Most have external inductors. What you could do is to split the tubes into two groups. 7.5 V, 2.5 A is more manageable than 7.5 V, 5 A.
Something like the LM2673 might be a good option. 3 A output so you'll need two. They only need a Schottky diode, an inductor, and a few inexpensive passives to run. Available in TO-263 that's easy to solder.
The LMZ1200x parts are rather spendy but they are so easy and bulletproof to work with. A handful of resistors, a couple of caps and you're done. The layout is super easy. But unfortunately, they only go up to 6 V on the output... So for 300B, 2A3 and a few others they're just about ideal.
As others have pointed out, there are options for linear regulators out there. You can either buy a 5 A regulator (and have no margin) or put a pass transistor on a smaller regulator (see the LM317 datasheet for examples). But why?! Once you're done doing the math at worse case conditions, you'll find that you end up burning as much power in the linear regulators as you do in the filaments. At least that was the conclusion I arrived at. Hence, I went with switchers...
~Tom
Tom,
I think you're advice is spot on. Those higher current regulators tend to get pretty complex and none seem to have simple to solder packages. I'll go with two smaller regulators. I thought it would simplify things to just do one but that doesn't seem to be the case.
Others,
Thanks for the advice, but I don't want to do a linear supply due to the wasted power. This design is going to get hot already and I don't want to add more heat than I need to. Also, soft start pin is a very attractive option pulling me toward the switchers. Plus, I wouldn't be learning anything if I go that route.
I think you're advice is spot on. Those higher current regulators tend to get pretty complex and none seem to have simple to solder packages. I'll go with two smaller regulators. I thought it would simplify things to just do one but that doesn't seem to be the case.
Others,
Thanks for the advice, but I don't want to do a linear supply due to the wasted power. This design is going to get hot already and I don't want to add more heat than I need to. Also, soft start pin is a very attractive option pulling me toward the switchers. Plus, I wouldn't be learning anything if I go that route.
Hi
I was searching the same topic and hope someone can reply on my question.
I am collecting parts to build 300B SET amp, and somehow I'm 'convinced' to go after smps tube heaters for 300B and driver, does it matter where I connect heater + and - from smps to 300B?? This question I raise, as all we know cathode is connected with heaters.
thanks
I was searching the same topic and hope someone can reply on my question.
I am collecting parts to build 300B SET amp, and somehow I'm 'convinced' to go after smps tube heaters for 300B and driver, does it matter where I connect heater + and - from smps to 300B?? This question I raise, as all we know cathode is connected with heaters.
thanks
I am using switchmode supplies for the filaments in my 300B amp. They work really well. I am selling assembled modules as well as bare boards for DIY use on my website (Neurochrome.com : : Audio : Universal Filament Regulator). These modules have been extensively tested, both in the lab and in my stereo in the living room.
Unless you are using several 300B tubes in parallel, you will need one regulator per 300B tube. For parallel tubes, you can run two tubes off of one of my filament regulators.
The (-) output of the filament regulator should be connected to the lowest potential of the 300B filament. This is generally the end where the cathode resistor is connected. If one end of the 300B filament is connected to ground, connect the (-) terminal of the filament regulator there. The (+) terminal then connects to the other end of the filament. I hope this makes sense...
I should draw up some figures of this at some point...
~Tom
Just a reminder - depending on their design, not all of the shelf DC-DC switching regulator solutions will provide isolation between input and output. If there is a cathode resistor present in the amplifier circuit, then the SMPS for the filament must be isolating, otherwise multiple supplies may act to link cathodes that should not be linked (between channels, for example).
Just wondering...have any of you switch-o-philes considered using a HV SMPS for the B+?
I'm working up a plan for an 833C power amp with 2300V B+
and was wondering if a switching option exists (such as those used for medical and scientific equipment) in addition to the usual linear supplies with 200lbs of iron and caps.
With a small lightweight PS I could easily get everything in one chassis...
I'm working up a plan for an 833C power amp with 2300V B+
and was wondering if a switching option exists (such as those used for medical and scientific equipment) in addition to the usual linear supplies with 200lbs of iron and caps.With a small lightweight PS I could easily get everything in one chassis...
Most of the DC-DC switchers are in fact not isolated between input and output. So in applications where the heater supplies are floating, you'll need separate regulators for each tube. Just as you would need separate transformer windings in a traditional supply.
In the case where one end of the heater/filament is at ground, the transformer winding, rectifier, and reservoir cap may be shared between regulators, but you'll still need a separate regulator for each DHT tube.
~Tom
Have a look at Post #9.
On a satellite project years ago, I developed a 2500 V flyback switchmode supply. Worked pretty well... It wasn't designed for high current so it wouldn't be as useful for a B+ regulator, but I would think it would be possible to build a B+ regulator that way.
It'd be quite a project, though. Switchers contain a lot of little trap doors and pitfalls -- especially the ones that handle high energy. Be careful.
~Tom
I wouldn't be building it, I'd be buying it. I see lots of companies that OEM these things for scientific instrument and medical scanners. Most have plenty of voltage but are a little short of current. I'd need 2300V and 160mA per monoblock, so that's ~370W. Most of the ones I see on the web are 200 - 250W. Now I suppose two could be paralleled to get enough wattage, but the question is - how would these things sound?? Have they solved the RF issues or would I be swamping my lovely tubes with all sorts of HF noise?
Just exploring the options at this point and trying to think outside the box.
Just exploring the options at this point and trying to think outside the box.
You will have some amount of ripple at the switching frequency. That's simple physics. But if the switching frequency is above the audio band and the ripple voltage is low, does it impact circuit performance? My suspicion is that you won't hear the ripple as it's well outside the audio band (my filament regulators run at 500 kHz - most mains level regulators run well above 20 kHz). And the ripple is tiny compared to the normal 120 Hz ripple you have on a linear supply - unless you design a linear voltage regulator. Switchers - or any series regulator - will have a very low output impedance. In my experience (from my 21st Century Maida Regulator), that really tightens up the sound. If you want a low noise floor, low hum, lots of sound stage go with a regulated supply. That's my subjective experience anyway.
In my case, I'm using switchers for the filament power. In that case, any switching noise would actually be amplified by the tube. I have not seen any adverse effects of using switchers on the filament supplies. The amp measures well and sounds good.
Buying a switcher for the B+ application sounds like a good idea. It's a bit safer that way. 2300 V, 160 mA. Damn. That's a lot of power.
~Tom
In my case, I'm using switchers for the filament power. In that case, any switching noise would actually be amplified by the tube. I have not seen any adverse effects of using switchers on the filament supplies. The amp measures well and sounds good.
Buying a switcher for the B+ application sounds like a good idea. It's a bit safer that way. 2300 V, 160 mA. Damn. That's a lot of power.
~Tom
Last edited:
Tom, thank you for the reply.
It makes sense connecting filaments as you suggested.
On my 300B amp project I'm collecting parts to build, will be only one tube on output stage, and 8 ohm speaker per channel.
I know this thread is for smps and tube heaters, but you are experienced on tube amps, and other question will be, what output transformer I should consider to purchase in terms of primary resistance? I can see 2.5K, 3.5K, 4.2 and 5K primary and this confuses me.. Regardless the design, does the output transformer resistance makes a big difference and why, what will be your decision, and if you have any suggestion or brand just let me know.
I like your modules which you are offering on your website, they could make my life easier..
thanks
bekim
Last edited:
Cool!
I think a better thread for this question is my 300B Amp build thread:
http://www.diyaudio.com/forums/tubes-valves/218217-deathtrap400-pretty-damn-good-300b-amp.html
I'll answer it there.
I like your modules which you are offering on your website, they could make my life easier..
Yep. That's why I started doing modules. That takes the stress of SMD soldering out of the equation for you.
~Tom
These units from Ultravolt (link) are oh so close, with the 2C units capable of 2000V and 125mA. Just shy of what I need. Output impedance is ~2ohms according to the applications engineer I talked to! Talk about a responsive power supply!
Too bad they cost $1300 each or I'd parallel two per monoblock and be done with it. I guess that's cheap compared to a WAVAC 833 amp ($350K!), but not according to my wife.
http://www.ultravolt.com/uv_docs/HPCDS.pdf
Too bad they cost $1300 each or I'd parallel two per monoblock and be done with it. I guess that's cheap compared to a WAVAC 833 amp ($350K!), but not according to my wife.
http://www.ultravolt.com/uv_docs/HPCDS.pdf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.