An externally hosted image should be here but it was not working when we last tested it.
That's right, I'm building a tube oscilloscope, with a switching power supply.
It doesn't *do* anything right now. The HV works (hence the green dot), and the heaters work, but no circuit is hooked up.
A lot of people might be interested in the power supply, by the way. Sans HV generator, it's perfectly suited for the average tube amp. The main outputs are +/-250V 200mA (which could be rearranged for +500/+250) and 6.3VDC 10A, both filtered and regulated. Residual ripple under light load is on the order of 0.1Vp-p (120Hz and 100kHz), the former of which is significantly lower than almost any conventional tube circuit, and the latter of which is very easy to filter out.
Tim
Iv'e been running switching power supplies in two of my tube amps ("Shrine" and "Mighty Mite") for some time now with good results (and small size), and am working on several others. Having said that, I spent a lot of time in the EMI room at my job optimizing those supplies for low emissions. It wasn't easy, as the HV output winding is just as bad a source of noise as the primary winding.
The HV winding waveform is ugly as sin. 
Lots of ringing from the flat wound construction -- didn't have room for bank winding, which I did for the filter inductor (choke input filter, it's a forward converter). It doesn't seem to be going very far thanks to the aluminum panels, and there doesn't seem to be much HF trash beyond the ~20MHz ringing of said transformer, at least on the scope.
I've taken adequate precautions with filtering, common mode and otherwise, so ground loop should be no problem at all. What's more, the voltages all come out on terminal strips, so I can nip any HF noise in the bud without it going places.
Tim
Lots of ringing from the flat wound construction -- didn't have room for bank winding, which I did for the filter inductor (choke input filter, it's a forward converter). It doesn't seem to be going very far thanks to the aluminum panels, and there doesn't seem to be much HF trash beyond the ~20MHz ringing of said transformer, at least on the scope.I've taken adequate precautions with filtering, common mode and otherwise, so ground loop should be no problem at all. What's more, the voltages all come out on terminal strips, so I can nip any HF noise in the bud without it going places.
Tim
A flyback converter is far more appropriate for HV applications than a forward, especially a ZVS flyback. 1) Since the flyback has boost capability as part of its transfer function, the transformer ratio need not be so extreme. 2) You get rid of the output choke 3) with ZVS flyback you operate in borderline discontinuous mode which greatly reduces the stress on, and noise from, output diodes. With ZVS flyback you also get the option of shaping both the leading and trailing edge of the primary waveform, further reducing EMI.
The high voltage generator has a lovely waveform, a piecewise cross between sine, tilted triangle and square waves, plus a small amount of ringing. It's a blocking oscillator in the flyback domain, but the HV winding has enough capacitance that the rising/falling edges are quite gradual, and it operates in continuous rather than single-cycle (or burst / squegging) mode. The HV output is a doubler, which flattens the top and bottom peaks.
Flyback supplies are attractive for lower power levels. Forward converters start looking good around 100W, which is the size of this power supply. I don't have any BCM flyback chips, so it would've been with UC3842, which can be far noisier, especially if it enters CCM. So I went with forward, which has lower ripple current, useful around the high ESR HV electrolytics and giving an easier time for the high amperage heater winding, at the expense of a somewhat more complicated HV filter choke.
I am considering PFC + flyback for a SS scope. That PSU might end up a fairly large board: it needs filter, PFC, flyback, more filters, HV generator (Royer oscillator, ran from LV supply), and still more filtering.
Tim
Flyback supplies are attractive for lower power levels. Forward converters start looking good around 100W, which is the size of this power supply. I don't have any BCM flyback chips, so it would've been with UC3842, which can be far noisier, especially if it enters CCM. So I went with forward, which has lower ripple current, useful around the high ESR HV electrolytics and giving an easier time for the high amperage heater winding, at the expense of a somewhat more complicated HV filter choke.
I am considering PFC + flyback for a SS scope. That PSU might end up a fairly large board: it needs filter, PFC, flyback, more filters, HV generator (Royer oscillator, ran from LV supply), and still more filtering.
Tim
I'd be tempted to use a high side power factor correction chip
to get boost regulated +390. Then chop 50%/50% duty into a
push-pull isolation transformer. Doubt I'd bother with any loop
feedback, let back EMF into the primary PFC take care of it...
Oh, Royer... I just looked that up. We on same sheet of music.
http://wiki.4hv.org/index.php/Royer_oscillator
Thinking 6080 here... Too much plate voltage? Too much Rpc?
Miller? Need separate switcher for isolated/unisolated filaments...
Maybe sand is better.
to get boost regulated +390. Then chop 50%/50% duty into a
push-pull isolation transformer. Doubt I'd bother with any loop
feedback, let back EMF into the primary PFC take care of it...
Oh, Royer... I just looked that up. We on same sheet of music.
http://wiki.4hv.org/index.php/Royer_oscillator
Thinking 6080 here... Too much plate voltage? Too much Rpc?
Miller? Need separate switcher for isolated/unisolated filaments...
Maybe sand is better.
Last edited:
The widemouth frog was going along one day and he met a wabbit.
'Hello, furry creature with the long ears and little white stumpy tail, what kind of a creature are you?'
And the wabbit said 'Oh, I'm a wabbit, I eat mostly grass, although I'm coprophagous and crepuscular to boot'
'Oh, right' said the widemouth frog, marvelling at these long words and smiling broadly 'I'm a widemouth frog, I eat flies and other insects, nice to meetcha'.
A little further on the widemouth frog came across a fox.
'Oh, swift creature with the red fur and pointy ears and impressive bushy tail, what kind of a creature are you?' said the widemouth frog.
'Oh' said the fox, 'I'm a fox, I run around staying out of the way of hounds, and among other things I eat wabbits, when I can spot the crepuscular little sods in the twilight. What kind of a creature are you?'
'Oh' said the widemouth frog smiling broadly in relief that he wasn't a wabbit, since the fox was rather fierce, 'I'm a widemouth frog, I eat flies and other insects, nice to meetcha'.
A little further on the widemouth frog came across a green snake.
'Oh, long and slithery green creature' he said 'what kind of a creature are you?'
'Ahh... sss,' said the green snake, hissing after the fashion of snakes. 'I am a green ssssnake, and I eat widemouth frogssss.'
'Ew' said the widemouth frog, pursing his lips as narrow as they would go, 'You don't see many of those around here...'
w
'Hello, furry creature with the long ears and little white stumpy tail, what kind of a creature are you?'
And the wabbit said 'Oh, I'm a wabbit, I eat mostly grass, although I'm coprophagous and crepuscular to boot'
'Oh, right' said the widemouth frog, marvelling at these long words and smiling broadly 'I'm a widemouth frog, I eat flies and other insects, nice to meetcha'.
A little further on the widemouth frog came across a fox.
'Oh, swift creature with the red fur and pointy ears and impressive bushy tail, what kind of a creature are you?' said the widemouth frog.
'Oh' said the fox, 'I'm a fox, I run around staying out of the way of hounds, and among other things I eat wabbits, when I can spot the crepuscular little sods in the twilight. What kind of a creature are you?'
'Oh' said the widemouth frog smiling broadly in relief that he wasn't a wabbit, since the fox was rather fierce, 'I'm a widemouth frog, I eat flies and other insects, nice to meetcha'.
A little further on the widemouth frog came across a green snake.
'Oh, long and slithery green creature' he said 'what kind of a creature are you?'
'Ahh... sss,' said the green snake, hissing after the fashion of snakes. 'I am a green ssssnake, and I eat widemouth frogssss.'
'Ew' said the widemouth frog, pursing his lips as narrow as they would go, 'You don't see many of those around here...'
w
For high voltage output, flybacks are actually optimal at higher powers than the conventional 100W wisdom point. Large CRT-based color TVs are a case in point. All of them require 100W+ of 150-180V to run their horizontal deflection/HV section. This is the bulk of the power used by the TV. Every CRT color TV I've seen uses a discontinuous flyback or quasi-resonant flyback in the power supply for reasons of noise generation. By discontinuous I mean that the stored energy in the transformer discharges and the output diode turns off before the next switch cycle begins. This is in contrast to continuous operation, where the output diode is still conducting from the previous switching cycle when the next onwe commenses, forcing into into hard recovery.
In my job, we tried powering a medium-sized color TV using a continuous flyback because the ICs we were making at the time (integrated MOSFET/controller) could not handle the required power without using a continuous mode flyback. Try as we might, with the best output rectifiers at the time, we found it impossible to get rid of snow on the screen at marginal signal levels, due to the reverse recovery transient from the output rectifier. The conventional TV supplies were meeting this goal using discontinuous flybacks and 600V, 250ns recovery output diodes that you can get from just about anyone. This lesson was not lost on me, so when I tried building an SMPS for my first tube amp, I used an RCC (self-oscillating) flyback that always runs in borderline discontinuous mode. This was OK, though there were stil a lot of EMI battles to be fought because of the HV output winding. There was still "excess" noise due to the fact that the primary switch turns on with substantial voltage across it. Using a quasi-resonant flyback fixed that, as the primary switch drain voltage rings down almost to zero before it turns on.This plus the discontinuous mode operation, makes EMI much easier to control for a quasi-resonant flyback.
I plan to take this topology up to a couple of hundred watts to power some GM70 monoblocks. I'm at 60-70W at present. Primary switch current isn't a big problem these days, as the silicon for handling the peak current is cheap and readily available.
In my job, we tried powering a medium-sized color TV using a continuous flyback because the ICs we were making at the time (integrated MOSFET/controller) could not handle the required power without using a continuous mode flyback. Try as we might, with the best output rectifiers at the time, we found it impossible to get rid of snow on the screen at marginal signal levels, due to the reverse recovery transient from the output rectifier. The conventional TV supplies were meeting this goal using discontinuous flybacks and 600V, 250ns recovery output diodes that you can get from just about anyone. This lesson was not lost on me, so when I tried building an SMPS for my first tube amp, I used an RCC (self-oscillating) flyback that always runs in borderline discontinuous mode. This was OK, though there were stil a lot of EMI battles to be fought because of the HV output winding. There was still "excess" noise due to the fact that the primary switch turns on with substantial voltage across it. Using a quasi-resonant flyback fixed that, as the primary switch drain voltage rings down almost to zero before it turns on.This plus the discontinuous mode operation, makes EMI much easier to control for a quasi-resonant flyback.
I plan to take this topology up to a couple of hundred watts to power some GM70 monoblocks. I'm at 60-70W at present. Primary switch current isn't a big problem these days, as the silicon for handling the peak current is cheap and readily available.
Ah, but then you get full 120Hz ripple, which is almost as bad as a cap input filter of the same size. The voltage is stabilized, but it's not regulated as such. I can't tolerate that kind of stuff going into my oscilloscope, and most people wouldn't tolerate that going into their SET, either!
Forward converters are also quite nice after PFC, as the duty cycle range is much lower = less ripple current drawn. YMMV.
Well, I was going to use a pair of 2N4401 and 10V supply, since it's just a few watts. 2kV at under 1mA, and 6.3V 0.3A.
The waveforms are very smooth, little switching noise. The HV winding always has lots of capacitance, so it's easy to be self resonant. These are nice things, especially in an oscillscope.
These are salvaged from equipment. The trick is to heat it slowly to avoid thermally stressing the core, and hot enough to soften the varnish. The bobbin is also brittle, so proceed carefully.
If salvage isn't enough, there are a few suppliers, such as Amidon, CWS Bytemark, Adams Magnetic, etc.
Tim
Last edited:
120Hz ripple after PFC? Last one I fooled with took about +/-85VDC to kick
start. But I don't know if it drops out each and every half cycle. Maybe or
maybe not there would be some dead time in-between? I guess while running
on 390V in the can, there is no "regulation" provided by PFC.
PFC still a much less wide dead spot to ride through than standard rectification,
and you do at least know the voltage level that isn't deadtime will be consistant.
You suggesting I should PWM the Royer to further smooth out 120 ripple?
Or better to just clean up the entire aftermess with a linear reg or gyrator?
start. But I don't know if it drops out each and every half cycle. Maybe or
maybe not there would be some dead time in-between? I guess while running
on 390V in the can, there is no "regulation" provided by PFC.
PFC still a much less wide dead spot to ride through than standard rectification,
and you do at least know the voltage level that isn't deadtime will be consistant.
You suggesting I should PWM the Royer to further smooth out 120 ripple?
Or better to just clean up the entire aftermess with a linear reg or gyrator?
Last edited:
Instad of messing around with still another switching stage, I'm temped to add a choke to a doubler circuit for a partial PFC. This was done with a lot of PC power supplies from the major manufacturers before specs got tighter and active PFC became a necessity. The required choke is not all that large - I have a couple salvaged from old PC supplies. It'll hardly be noticed next to the output transformers...
Good question. I'm aiming for 5MHz, which is on par with the earliest Tek scopes, but taking advantage of high Gm tubes they didn't have in the late '40s, so I won't need their distributed amps. Tubes like the 7KY6 frame grid pentode, and the 6DK6 which Tek later used lots of. Both are shown in the picture -- this is enough for a simple X-Y plotter, an excellent start. The 6DK6s are used in pairs as diff amps, driving 7KY6 balanced deflection outputs. A 6DJ8 follower (one triode per channel) drives the diff amp input.
Later, I'll add preamps for additional sensitivity (for something around 10mV/div lowest range), fix up my own attenuator and stuff (the original B&K 1450's attenuator had a 3-6 series, how awful!), and build the sweep.
I'm considering a triggered delayed sweep, but it won't be as nice as a Tek's, keeping it fairly simple while providing adequate functionality. I'm expecting a 50ns/div top range, which will take 500ns to cover the screen, and will have plenty of minimum holdoff time. I haven't decided what kind of trigger element I want to use; if it's tube, it could be quite slow, and I'd like to have at least a 5MHz trigger bandwidth. That basically rules out a verbose comparator-and-logic implementation, but doesn't exclude a fast schmitt trigger. I do want to do as much as possible in tubes, which should prove interesting when it comes to generating constant currents in both directions!
Tim
Sch3mat1c,
Get yourself on the TEK scopes forum
I am sure the guys over there would be very pleased to see your project. There is no dought some of them could help you with trggered delay etc. Here it is:
TekScopes : All about classic Tektronix CRT o'scopes
They are the boys
Cheers Matt.
Get yourself on the TEK scopes forum
I am sure the guys over there would be very pleased to see your project. There is no dought some of them could help you with trggered delay etc. Here it is:
TekScopes : All about classic Tektronix CRT o'scopes
They are the boys
Cheers Matt.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.