I found this PS module at eBay. I am curious if anyone tried it.
https://www.ebay.ca/itm/36436657558...uid=uMxPcLkpRsm&widget_ver=artemis&media=COPY
https://www.ebay.ca/itm/36436657558...uid=uMxPcLkpRsm&widget_ver=artemis&media=COPY
Attachments
I got one, but I have not used it in a real amplifier yet. From the testing:
- The claimed 250W is off the mark. The transformer will get very hot with 250mA HV load.
- HV heavily depend on the 12.6V/6.3V current load. You need to have at least 2A load to be able to get closer to the claimed HV voltage.
I don´t like the idea that the HV is dependent on the heater current, it tells me that the regulation is only on the LV output. Swap tubes, i-e heater current and all HV havoc voltage prevails, No! unless after post-regulation is used which would seem a waste.
Regulation seems typical of a switching topology having poor cross-output regulation, and it looks there is no coupled filter inductor used on each switching output. That rules out forward topology. Also, on paper there is no emissions compliance regulation that interference produced could play havoc with nearby radio etc. Switchmodes are designed to run hot to limit core size, power losses roughly 50% ferrite and 50% copper. Running temp on full load 30°C over 25°Cambient isn´t unusual. That implies 55°C to touch, requiring min F or H grade UL listed insulated copper wire.
Of course only my view;
By all means proceed, but use an isolating transformer, but take care and don´t get stung. Keep loads connected !
Over the decades I have designed several 1kW PFC 450V 2A, versions and higher volts for tube amps but using a mains transformer for the heater supply. Also mention that any serious switchmode design requires dramatic knowledge and experience bumps, way upward of Tube amps. However, once the subject is mastered, there are negligable voltage droops on transients with excellent regulation benefits. There is also more to go wrong.
Viva SMPS
Bench Baron
Regulation seems typical of a switching topology having poor cross-output regulation, and it looks there is no coupled filter inductor used on each switching output. That rules out forward topology. Also, on paper there is no emissions compliance regulation that interference produced could play havoc with nearby radio etc. Switchmodes are designed to run hot to limit core size, power losses roughly 50% ferrite and 50% copper. Running temp on full load 30°C over 25°Cambient isn´t unusual. That implies 55°C to touch, requiring min F or H grade UL listed insulated copper wire.
Of course only my view;
By all means proceed, but use an isolating transformer, but take care and don´t get stung. Keep loads connected !
Over the decades I have designed several 1kW PFC 450V 2A, versions and higher volts for tube amps but using a mains transformer for the heater supply. Also mention that any serious switchmode design requires dramatic knowledge and experience bumps, way upward of Tube amps. However, once the subject is mastered, there are negligable voltage droops on transients with excellent regulation benefits. There is also more to go wrong.
Viva SMPS
Bench Baron
Amp enthusiast quite close to me put a low power 200W SMPS into an SE amp chassis with direct heater 300B and he had problems with switching harmonics entering the heater section. The interference levels were already fairly low, but incorrect chassis node earthing is is asking for problems, requiring suppression ferrites, cap/chokes that basically filled the sub chassis. One is warned ! Read those who have/had issues with just 50/60Hz let alone high frequency switcher harmonics. An indirect heater design would be a better solution but may not be the remedy.
It is often said that although switchers "weight to power ratio" is fantastic compared to iron lumps; by the time the noise and spikes are dealt with to make all voltages noise free, one ends up with a design that fills the same or more space an iron transformer does with a big cover! Tread warily.
Bench Baron
It is often said that although switchers "weight to power ratio" is fantastic compared to iron lumps; by the time the noise and spikes are dealt with to make all voltages noise free, one ends up with a design that fills the same or more space an iron transformer does with a big cover! Tread warily.
Bench Baron
One thing to keep in mind with switch mode power supplies is the peak to RMS (average) power ratio drawn from the power supply by a class ABx push pull amplifier. The B+ part of the switch mode power supply has the be sized more than two times the power output to be delivered by the amplifier.
For example.
Neglecting output stage losses to deliver 50 watts RMS from a class AB amplifier the power supply has to supply root 2 squared of power to the amplifier.
So before amplifier losses are added the power supply needs to deliver 100 watts peak for a 50W RMS class AB amplifier.
In 50/60Hs power supplies this peak power comes from the filter capacitors and some power supply output voltage droop.
This is not usually the case for switch mode power supplies.
Many switch mode power supplies use small amounts to total capacitance in the power output reconstruction circuits as the switch frequency is very high (>40Khz) so the output capacitors only need to store energy for milliseconds not a fraction of a second like a 60Hz power supply.
As a result the output capacitors will be quickly discharged by long low frequency peak currents and the output voltage may start to quickly droop.
To make this worse the power supply may go into over current protection resulting in the output voltage dropping off even more quickly.
This becomes progressively more important as the frequency amplified by the class AB amp is lowered. So you may be OK at 1Khz but find the switcher supply output voltage drops greatly at 30Hz. Been there, been caught.
So when sizing the switcher for the B+ load be very conservative. Start with 2 times the class AB amplifier power output and then add margin for amplifier losses.
Allowing for 3 times the amplifier output power in peak switcher supply B+ power rating is not too conservative.
Of course the above de-rating does not apply to any heater load.
However if it is a china power supply de-rating all continuous loads by at least 50% is likely going to be a very good idea if long life is to be expected.
The example china supply posted looks like it would be good with 6BQ5 or so sized outputs for a stereo amplifer.
And yes noise is likely to be a issue to be solved so be prepared to chase down the switcher noises in your amplifier.
A switcher can however work very well and allow the construction of smaller and less hot amplifiers.
For me I mostly go with big iron transformers, chokes and tube rectifiers for the "romance" of it all. LOL
For example.
Neglecting output stage losses to deliver 50 watts RMS from a class AB amplifier the power supply has to supply root 2 squared of power to the amplifier.
So before amplifier losses are added the power supply needs to deliver 100 watts peak for a 50W RMS class AB amplifier.
In 50/60Hs power supplies this peak power comes from the filter capacitors and some power supply output voltage droop.
This is not usually the case for switch mode power supplies.
Many switch mode power supplies use small amounts to total capacitance in the power output reconstruction circuits as the switch frequency is very high (>40Khz) so the output capacitors only need to store energy for milliseconds not a fraction of a second like a 60Hz power supply.
As a result the output capacitors will be quickly discharged by long low frequency peak currents and the output voltage may start to quickly droop.
To make this worse the power supply may go into over current protection resulting in the output voltage dropping off even more quickly.
This becomes progressively more important as the frequency amplified by the class AB amp is lowered. So you may be OK at 1Khz but find the switcher supply output voltage drops greatly at 30Hz. Been there, been caught.
So when sizing the switcher for the B+ load be very conservative. Start with 2 times the class AB amplifier power output and then add margin for amplifier losses.
Allowing for 3 times the amplifier output power in peak switcher supply B+ power rating is not too conservative.
Of course the above de-rating does not apply to any heater load.
However if it is a china power supply de-rating all continuous loads by at least 50% is likely going to be a very good idea if long life is to be expected.
The example china supply posted looks like it would be good with 6BQ5 or so sized outputs for a stereo amplifer.
And yes noise is likely to be a issue to be solved so be prepared to chase down the switcher noises in your amplifier.
A switcher can however work very well and allow the construction of smaller and less hot amplifiers.
For me I mostly go with big iron transformers, chokes and tube rectifiers for the "romance" of it all. LOL
Last edited:
Good sum up.....I get around the peak PSU current by using a boost mode power factor IC (older L4981A by ST) and that emphasises an appropriately sized output bulk cap that nicely rounds up the ripple issue. That makes the input fuse rating much more appropriate. As with most buck step up power topologies which nicely fit 2 x AC EU input for 400V B+, much of the noise is reflected to the AC input. Again, another tactic I use is to deliberately oversize the switching inductor so that the ripple current is roughly 15% or less of main current, way less than commercial 30%. That makes suppression easier to deal with. The other tack is experience in obtaining the correct ferrites for suppression, amorphous toroids and others are an absolute must.
However, for the newcomer (I apologize for the complexity) all this seems a minefield and is unfortunately no easy way. We haven´t yet dealt with mosfet/diode tab charge-coupled interference though the heatsink, this is another trick to double insulate the heat transfer washers by slipping a piece of copper foil in between which is directly earthed to the switching earth. This earth is NOT the main earth !! but a return for RF switching bypass currents.
When doing signal to noise measurements, it is most likely some switching harmonics will leach through, and exquisit RF and microwave screening techniques will work. Be prepared for a midnight plumbing exercise.
Welcome to the kW area.!
Bench Baron
However, for the newcomer (I apologize for the complexity) all this seems a minefield and is unfortunately no easy way. We haven´t yet dealt with mosfet/diode tab charge-coupled interference though the heatsink, this is another trick to double insulate the heat transfer washers by slipping a piece of copper foil in between which is directly earthed to the switching earth. This earth is NOT the main earth !! but a return for RF switching bypass currents.
When doing signal to noise measurements, it is most likely some switching harmonics will leach through, and exquisit RF and microwave screening techniques will work. Be prepared for a midnight plumbing exercise.
Welcome to the kW area.!
Bench Baron
Here is the underchassis pic of the final assembly of a large stereo parallel configured KT90 HiFi power amp, also designed for bridge mode operation to provide around 400W output. With the screening lid off the boost mode converter, this switching topology provides 450V at 2.5A.
The give-away area is the toroid chokes and the large square main inductor. This design is a step down from a version with even higher power 500W using 650V B+ which annoyingly required stacked electrolytics. Note the tight switching area layout, as hard switching is used with ground planes within "ground planes". Both amps provide HiFi spec and low distortion from 15Hz to 40Khz.
The caveat....A veritable amount of bench test gear is required with both digital and analog scopes.
Bench Baron
The give-away area is the toroid chokes and the large square main inductor. This design is a step down from a version with even higher power 500W using 650V B+ which annoyingly required stacked electrolytics. Note the tight switching area layout, as hard switching is used with ground planes within "ground planes". Both amps provide HiFi spec and low distortion from 15Hz to 40Khz.
The caveat....A veritable amount of bench test gear is required with both digital and analog scopes.
Bench Baron
Attachments
In addition to the usual conducted and radiated interference, using a mains → DC converter to power the filament of directly heated tubes (300B) can give other troubles.
for example. every such switching converter has a leakage current (with wideband noise) that tries to reach ground. If connected to the 300B filament, the leakage current partly runs through the filament - which is the same conductor for the anode current.
The noise conducted from the SMPS converter into the mains network is another problem. The EMC regulations like EN55022/CISPR22 do not care about noise at 150kHz or lower, and the noise in this band is usually very high. This noise can cause trouble for your DAC, or DAC oscillator, or circuits with JFETS, or older opamps that are not "EMI-hard"
I have eyes on these for a while: they would make a nice bench supply for messing around, and/or a spare part for my tube tester, if I could figure out a way to make the HV variable.
.
As for listening, no Sirs, no way: I don't allow any SMPS in the signal path; even my DAC has linear PS. And yes, the PS is in the signal path.
.
As for listening, no Sirs, no way: I don't allow any SMPS in the signal path; even my DAC has linear PS. And yes, the PS is in the signal path.
Unfortunately the proliferation of SS switching noise emission is getting worse in the mid-upper RF band, (again back to the fella who tried the switcher with the SE amp = failed attempt) also transpires his roof solar converter was also a heavy mains noise contribant; ....solution here, see pi. With the impossibility of fitting ferrite chokes within the converter chassis screen, the temporary bodge-up mounted externally worked, so he can listen to radio without interfering "birdies and whistles". One wonders how many commercial units spew unchecked "trash" out, and claim, as you mention EN/CISP approvals ? I suspect alot of lies in the trade to save costs. These chokes are quite cheap, example typical Mouser part 623-2631102002. The trick is to know which frequency/noise is creating mayhem and a spectrum analyser or sim instrument is vital.
Maybe at this juncture, the forum moderators may decide to throw us to the sand section or start a new thread as we have deviated somewhat from the true dealings of Tube amp PSU´s. However, at least the switcher issues have been aired and can be judged by others who don´t have the experience and remain the wiser !
I did my SState best with tram, escalator & train motive power drives, inconsideration of sitting public only meters away from 10´s of kW´s of variable frequency semiconductor drives, and yet can use their phones and notebooks with ease.
It can be done......but......Viva SMPS !
Bench Baron
Attachments
Just avoid this one:
High power gallbladder switch power supply KT88 300B electronic tube power amplifier stabilized voltage can be customized
It does actually provide all the supplies required for a stereo SE 300B amp and it was useful to get my amp up and running. It's an LLC resonant supply but only the HT is regulated. It's very well put together with plenty of mains filtering on the front end, common mode choke etc. I have a reverse engineered schematic if anyone is interested although the controller is on a opaquely-coated daughter board.
Rectified mains filtering is only 3 parallel 150uF capacitors. The resulting 100Hz ripple goes straight through the converter. Lots of ~90kHz switching noise on all supplies and weirdly loads of noise in the audio band. Maybe the controller is going into some sort of burst mode.
I was using Rod's filament regulators so maybe some of what Rod mentioned above was going on too. Currently rebuilding the amp with linear supplies.
That said, it has piqued my interest in building my own SMPS. Currently working on the PFC part. LLC resonant supply is too difficult though so will go for something else.
Comments above from benchbaron and BlueSystems are very interesting. It's a fascinating area.
High power gallbladder switch power supply KT88 300B electronic tube power amplifier stabilized voltage can be customized
It does actually provide all the supplies required for a stereo SE 300B amp and it was useful to get my amp up and running. It's an LLC resonant supply but only the HT is regulated. It's very well put together with plenty of mains filtering on the front end, common mode choke etc. I have a reverse engineered schematic if anyone is interested although the controller is on a opaquely-coated daughter board.
Rectified mains filtering is only 3 parallel 150uF capacitors. The resulting 100Hz ripple goes straight through the converter. Lots of ~90kHz switching noise on all supplies and weirdly loads of noise in the audio band. Maybe the controller is going into some sort of burst mode.
I was using Rod's filament regulators so maybe some of what Rod mentioned above was going on too. Currently rebuilding the amp with linear supplies.
That said, it has piqued my interest in building my own SMPS. Currently working on the PFC part. LLC resonant supply is too difficult though so will go for something else.
Comments above from benchbaron and BlueSystems are very interesting. It's a fascinating area.
I am using a 20v 4a sms for the heater supply on my X10 preamp but it feeds a LM350T regulator which drops it down to 13v (two PCC88s (7v heaters) in series.
When I experimented with a 12v one, I found that addditional filtering in the form of some Tants and a common mode choke worked well.
Switching frequencies seem to vary quite widely.
I wouldn't use one for the HT.
When I experimented with a 12v one, I found that addditional filtering in the form of some Tants and a common mode choke worked well.
Switching frequencies seem to vary quite widely.
I wouldn't use one for the HT.
......Well, so much for that idea.....then the noise fog appearing on the signal to noise ratio measurement with an "A weighting" audio bandpass filter will jarr all up.
I don´t want to hear any subharmonics within this important passband. It seems any attempt to use one of these switching power supplies for a CCS voltage source (a requirment for absolute zero noise) looks like a commencement for problems.
Seems these SS supply manufacturers (FE) deliberately don´t mention noise and ripple because it is poor to terrrible. Compared to the traditional tube amps of previous generations, the hum & noise ranged from -55 to -80dB down with usually soft low hum and perceptable hiss just discernable in the speaker/tweeter. At least the hum potentiomenter across the heater supply often provided an effective null but don´t assume the same for Switching PSU´s.
.....and it goes on.
Bench Baron
Wow! I just enlarged the picture and had a good look. You had to go to a lot of trouble to suppress noise and interference. Your comment about test equipment is noted.
I'm planning on PFC, LLC resonant converter for the HT and an isolated flyback driving buck or SEPIC regulators for filaments, heaters and bias. I've got as far as designing and laying-out the PFC. Still digesting all the SMPS PCB layout guidelines I can find.
I've built enough amps and don't mind spending a few years messing around with SMPS. I'll invest in an isolation transformer before I start testing the PFC.
Have a go, head in the pit job! ....Any new design I would go the same way, however the caveats on using resonant topology are just as valid for all Buck topologies, that is peak currents tend to be reflected to the primary side. There are benefits of a central freq clock to sync other modules, so long one is aware which one is producing excess harmonics.
At the time in the late 1970´s out of Uni I was snapped up by military companies in system electronics and followed much work by Lloyd Dixon of then Unitrode and Laszlo Balogh (current doubler worth a read), articles and manuals, now copyright Texas instruments www.ti.com, are a mine of information. Don´t forget general aspects of SMPS by Keith Billings too, but bit past the time. as SMPS principles have moved on. Surface mount is very attractive proposition but I have to resort to dentist lenses to get a move on. (Slow strained progress)
I would admit at that time around 1970 I had excellent going getting ferrites and samples of all sorts was dirt easy, from rare and excellent amorphous types by Allied Signal and others. However, quality Semi´s are better now, but much of that ease has gone unless one is part of an R&D department.
That power amp is a rebuild and modest step down from an even higher power version, but as mentioned running 450VB+ using 500V electrolytics to save doubling-up space. SMPS are very unforgiving on very low output impedances but soon shall use a spectrum analyser to see input and output harmonics.
Bench Baron
At the time in the late 1970´s out of Uni I was snapped up by military companies in system electronics and followed much work by Lloyd Dixon of then Unitrode and Laszlo Balogh (current doubler worth a read), articles and manuals, now copyright Texas instruments www.ti.com, are a mine of information. Don´t forget general aspects of SMPS by Keith Billings too, but bit past the time. as SMPS principles have moved on. Surface mount is very attractive proposition but I have to resort to dentist lenses to get a move on. (Slow strained progress)
I would admit at that time around 1970 I had excellent going getting ferrites and samples of all sorts was dirt easy, from rare and excellent amorphous types by Allied Signal and others. However, quality Semi´s are better now, but much of that ease has gone unless one is part of an R&D department.
That power amp is a rebuild and modest step down from an even higher power version, but as mentioned running 450VB+ using 500V electrolytics to save doubling-up space. SMPS are very unforgiving on very low output impedances but soon shall use a spectrum analyser to see input and output harmonics.
Bench Baron
Nice bit of design and construction you have there. That is a big switch mode transformer and they can be a real challenge.
I am always impressed with the quality of work I see in this forum.
Do you ever use Litz wire. I found it very good at reducing wire loss effects at higher frequencies like in resonant mode supplies.
I also found it very effective replacing any single thick wire in the transformer with several/many bifilar wound smaller wires in keeping wire loss and leakage inductance down.
Litz wire yes, and fairly spaced on homemade former. The core 3x EE42 pairs (with 1.5mm gap) is deliberately oversized to reduce reflected current ripple back on to mains, so the AC input side there is a pi-filter using amorphous chokes. Even though this version is hard switched, (tight layout crutial here, avoid PCB and go direct leg to leg to reduce inductance) the output of boost/buck class has much lower ripple, which is 2f and low amounts of switching harmonics. The PHE 470uF 500V cap has a brute low esr and any failings with both mosfet/diode switchers, both will explode off the board. The power supply has to manage a near 1:2 current ratio from quiescent to full load so I gave the core the size benefit though 2xE42 pairs would have sufficed. Many have asked me where to find quick formulation for magnetics/ core sizes and this quickie on the web provides a rough assessment. There are always caveats and trading specs with SMPS
Generally I fudge much, as I have quite alot of certain cores and the E cores I have can be paired up paralled and gap ground. The important parameter is core area and enough space for the wire. I tried the ETD large sizes and found magnetic radiation with a screening enclosure still an issue.
BB
Generally I fudge much, as I have quite alot of certain cores and the E cores I have can be paired up paralled and gap ground. The important parameter is core area and enough space for the wire. I tried the ETD large sizes and found magnetic radiation with a screening enclosure still an issue.
BB