Author Topic: SMPS for dormant tube audio amp project  (Read 2278 times)

0 Members and 1 Guest are viewing this topic.

Offline GKTopic starter

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
SMPS for dormant tube audio amp project
« on: August 19, 2019, 01:48:04 pm »
http://www.glensstuff.com/200vbuck/200vbuck.htm

The above is a link to a description of a mains-powered buck regulator I built nearly 10 years ago as part of a power supply unit for an audio amplifier project that has been sitting uncompleted on shelf since (which I'm revisiting and finishing off). This was to serve as a preregulator for a simple unregulated converter providing the 800V plate supply as well as +/-140V rails for the solid-state driver circuitry (the latter cleaned up and dropped down to +/-120V by a linear reg.

Performance wise the old design is still plenty adequate for the job at hand, but I'm contemplating perhaps redesigning the whole shebang and shrinking it all down to a single PCB if possible.

The IGBTs I used are obsolete now and there are probably better off-line half-bridge driver ICs available now too. At the moment I'm interested in how much higher I could practically raise the switching frequency to so as to shrink down my magnetics. For off-line voltage levels at this kind of power level, using more modern MOSFETs/IGBT/driver ICs, what is considered decent or run of the mill switching frequency wise in 2019?
« Last Edit: August 19, 2019, 01:58:43 pm by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline GKTopic starter

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #1 on: August 19, 2019, 01:55:19 pm »
Incidentally the buck regulators for the heater supplies with valves (just mocked up on a spare chassis for testing/evaluation) - low ripple, 60s linear voltage-ramping soft-start, crow-bar OVP, etc.



« Last Edit: August 19, 2019, 01:59:18 pm by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: SMPS for dormant tube audio amp project
« Reply #2 on: August 19, 2019, 05:37:36 pm »
Nice!

Hm, IGBTs were an odd choice; the high side fine, but the low side wasn't doing anything.

These days, anything up to 400kHz would be pretty normal I guess, and you can always go higher with GaN if you feel like it!

Schottky diodes are available in all voltages now, from 20V to 100V (conventional, Pt-Si I think, guard ring), to 200-300V (field-effect "super barrier" and superjunction), to 600V+ (SiC).  The only advantages held by fast PN diodes are cost and leakage (and maybe surge current).

MOSFETs, especially higher voltage types, are far better performance than they used to be.  Performance of all types has continued to improve incrementally; high-Vds parts have improved sharply (formerly, a quadratic performance blow at high Vds; superjunction removed this limitation).  There's also SiC and GaN, with different Vgs requirements; SiC has better Rds, and somewhat better speed, at high voltage ratings; GaN has crazy speed and is recently available in higher voltages (500V) as well as low voltages (30-200V).

If you want to keep the synchronous converter, shop for MOSFETs with Rds(ON) < 0.5V / Iout(max) or thereabouts.  That way you avoid body diode conduction (except at switching edges).  This is harder to do at higher and higher voltages, as you need quadratically beefier transistors, incurring significant cost and Qg.

Regarding switching edges: consider using as marginal a dead time as possible.  It only takes 10s, 100s of ns for the body diode to get forward-biased, and then you incur full hard-switching recovery and loss.  Some interleave isn't even a bad thing, as long as the switching loop has a controlled impedance (i.e., you may want to add stray inductance and a snubber).  The key insight is that, with controlled impedances, you aren't working against a steep cliff; rather, the switching losses increase gradually to either side, and you want to ride the valley of soft switching inbetween.

Dead time could even be controlled based on load current or switching voltage, which, I don't know of any controllers that attempt to do that, but an adventurous analog type like yourself might..?

Wait, I have seen some controllers that do that, but only sort of; the one example was a resonant controller, which extends dead time under conditions expected to have low load current (i.e., high switching frequency).  But still, nothing full-on watching load current, or commutation voltage, and operating accordingly.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline GKTopic starter

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #3 on: August 20, 2019, 07:37:59 am »
The low-side IGBT conducts in the forward direction when the load current drops below the peak ripple current of the inductor - it's there to ensure that the regulator always operates in continuous current mode and is how the control loop is made to operate entirely unperturbed right down to a load current of zero.

A MOSFET with low rds-on would give less conduction losses when conducting in the reverse direction than a IGBT here, but given the low average load current relative to the peak and the fact that the low-side switch is only on for ~40% of the time, the efficiency gains I think wouldn't go beyond a single watt. But either way I'm not fussed - MOSFET or IGBT - whatever gets the job done.

I have stock of all sizes of ETD formers and cores in F48 material, which is said to be good to 500kHz, but is in reality is getting a bit lossy by then as far as I can tell. ~200kHz or so might be a good target/all-round compromise.
« Last Edit: August 20, 2019, 07:55:31 am by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline GKTopic starter

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #4 on: August 20, 2019, 07:50:17 am »
Does anyone here have experience in efficiently grinding ferrite? This stuff is frigging hard and you can forget about a hand file! So far the best thing I have found for the job is these little abrasive flap wheels, but it's still a slow and tedious process.

I'm going to potentially have a mini hobby production run of direct off-line flyback transformers here soon which will need to be gapped, and I'm not looking forward to it.



« Last Edit: August 20, 2019, 07:52:29 am by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: SMPS for dormant tube audio amp project
« Reply #5 on: August 20, 2019, 11:36:09 pm »
Ah right, for continuous current, that'll do. :-+

Gapping -- if shims won't do (increased EMI?), should be okay with some carbide or diamond tooling, slow feeds and lots of water?

Suppliers will grind to order, but that may be a hundred thousands thing, not a hundreds thing?  Well, first things first I guess, were these ordered from an electronics distributor, or magnetics supplier?

Local alternatives, consider asking anyone that does grinding, stone/masonry/tile grinding, or even lapidary.  The former may object if they're only set up for metal grinding (fair enough), and the latter may not be consistent, or too labor-intensive (i.e., doesn't have a mill or surface grinder?).

Flap wheel, good effort, but dear god the mess that must've made? :-DD

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: Someone

Offline Circlotron

  • Super Contributor
  • ***
  • Posts: 3362
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #6 on: August 21, 2019, 12:08:35 am »
And with all this stuff in the local news about guys getting silicosis from cutting stone bench tops, I’d be careful about controlling the dust if you grind it yourself. Particulates of any sort in the lungs doesn’t sound nice.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: SMPS for dormant tube audio amp project
« Reply #7 on: August 21, 2019, 12:40:51 am »
And with all this stuff in the local news about guys getting silicosis from cutting stone bench tops, I’d be careful about controlling the dust if you grind it yourself. Particulates of any sort in the lungs doesn’t sound nice.

Hm, good point, I wonder if one can get manganism (a Parkinsons-like disease more often found in weldors) from such dust, or metal (usually zinc) fume fever (for those susceptible to it)?  Or nickel related diseases (dermatitis, or its internal equivalents; carcinogenicity, etc.).  Since, the ferrites in question are MnZn ferrite, but NiZn ferrites are also in use (for the RF range).  I at least wouldn't expect silicosis exactly; ferrite is probably dissolved or removed pretty easily from the lungs, it's what it does while it's there that's the question.

Ed: this fairly old paper seems to find it's symptomatic, but fairly innocuous; they may not've tested for further symptoms (e.g. neurological) though?
https://www.jstage.jst.go.jp/article/indhealth1963/21/1/21_1_1/_pdf
So that's not terrible news, and almost certainly not an acute exposure risk... but, I mean, if nothing else, I'd rather not do it without PPE, just because I'll look like a coal miner when I'm done. :o

Tim
« Last Edit: August 21, 2019, 12:47:59 am by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline GKTopic starter

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #8 on: August 21, 2019, 10:04:47 am »
I'm not really bothered by the dust. I really isn't that bad or hard to avoid breathing in. You should try power brushing/grinding a complete rusty 50 yo Land Rover chassis back to bare gleaming metal. I had infection (clogged pores turning into zits is my completely unqualified medical guess) problems inside my ear canals for ~two years after that just due to the sheer volume of ultra-fine dust and crap that managed to get all of the way into them. Having since discovered the amazing work that bead blasters can do I'm never going to do anything like that again.
 
There is no economic case for employing someone else to do the dirty work for just some hobby stuff which could be knocked off in just a single solid afternoon. I'm really just interested if there is some kind of proven alternative product to make the job easier that I could just pick up at a hardware store to or tool shop.
« Last Edit: August 21, 2019, 10:06:46 am by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: SMPS for dormant tube audio amp project
« Reply #9 on: August 21, 2019, 04:33:54 pm »
The low-side IGBT conducts in the forward direction when the load current drops below the peak ripple current of the inductor - it's there to ensure that the regulator always operates in continuous current mode and is how the control loop is made to operate entirely unperturbed right down to a load current of zero.

Yes, a synchronous buck operates in CCM over the entire load range, but current flows in the reverse direction through the low side switch during the freewheeling period, so you can't use an IGBT here. The circuit still works because of the co-pack diode, but you aren't actually getting any of the benefits (or downsides) of forced CCM.


Does anyone here have experience in efficiently grinding ferrite?

Yep, and it requires diamond to do so with any kind of efficiency. Solid carbide can work for a single use, but you'll trash the cutter in the process.

Note, however, that you can buy ETD cores pre-gapped even from the "mass market" distributors like Mouser and DigiKey. You're stuck with only certain gap (or AL) values, but there's usually a fair amount of flexibility in the buck choke [edit - or a flyback transformer], anyway (especially in forced CCM).

« Last Edit: August 21, 2019, 04:35:46 pm by MagicSmoker »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: SMPS for dormant tube audio amp project
« Reply #10 on: August 21, 2019, 06:15:55 pm »
Ah, well then... uh, tile saw?  Belt or barrel sander?  Dremel I don't think is going to be any easier than your existing flap-wheel solution, too small and tedious...

If you already have saws or grinders, a grinding wheel will do the job, and maybe you can make a little jig to hold the work at a fixed distance.  Drip on water in such a way that it doesn't get into the motor or create a shock hazard, and that should be okay...

Carbide should be pretty okay, ferrite isn't as hard or as strong as glass.  Would be more worried I think that you need a setup to make it repeatable.

Or, heck, if you already have an X-Y table, and a Dremel or the like, that'd be pretty sweet...

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: SMPS for dormant tube audio amp project
« Reply #11 on: August 21, 2019, 07:38:44 pm »
...
Carbide should be pretty okay, ferrite isn't as hard or as strong as glass. 
...

The problem with using a carbide cutting tool on ferrite is that they are both very hard and very brittle, so both are liable to chip or even shatter when they come into contact. A diamond-coated rotary burr* either in a Dremel or chucked up in a drill press (with a cross-slide vise) or a mill works far better.

FWIW, ferrite is gapped commercially with a diamond abrasive wheel in a surface grinder, which isn't something even the above-average home shop is likely to have. The only cutting process I have ever seen used on ferrite is a waterjet with carborundum grit.


EDIT - like this: https://smile.amazon.com/Diamond-coated-Cylindrical-Grinding-Mounted/dp/B015PK0IK6/ref=sr_1_30?keywords=diamond+burr&qid=1566414296&s=gateway&sr=8-30
 

Offline Circlotron

  • Super Contributor
  • ***
  • Posts: 3362
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #12 on: August 21, 2019, 11:32:22 pm »
The low-side IGBT conducts in the forward direction when the load current drops below the peak ripple current of the inductor - it's there to ensure that the regulator always operates in continuous current mode and is how the control loop is made to operate entirely unperturbed right down to a load current of zero.

Yes, a synchronous buck operates in CCM over the entire load range, but current flows in the reverse direction through the low side switch during the freewheeling period, so you can't use an IGBT here. The circuit still works because of the co-pack diode, but you aren't actually getting any of the benefits (or downsides) of forced CCM.
Yeah, that's a really good point. The lower IGBT is only doing something useful in DCM - keeping pulse width sensible at low load *and* providing ZVS for the upper IGBT. In that situation the filter choke current is bi-directional. When you get to CCM the choke current is uni-directional and as you say, during the freewheel period the IGBT won't conduct in the reverse direction, only the co-pack diode does, with it's attendant voltage drop. A mosfet can conduct in both directions, and during the freewheel period it shunts current around it's own body diode. So mosfet is a much better solution.
« Last Edit: August 21, 2019, 11:35:04 pm by Circlotron »
 

Offline GKTopic starter

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #13 on: August 22, 2019, 01:12:14 am »
Will you guys please read my second post in this thread
I forced ccm with the low side igbt entirely for the control properties  ; constant pwm d/c at all load currents

Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline Circlotron

  • Super Contributor
  • ***
  • Posts: 3362
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #14 on: August 22, 2019, 01:25:37 am »
Commenting in a forum can be a bit of a problem sometimes. I'm not criticizing your design, rather it's just a general comment for the sake of others reading that might not be totally up to speed on the finer points of switching regulators. No offense intended.
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: SMPS for dormant tube audio amp project
« Reply #15 on: August 22, 2019, 11:46:42 am »
Will you guys please read my second post in this thread
I forced ccm with the low side igbt entirely for the control properties  ; constant pwm d/c at all load currents

Hmm, rereading my previous reply I see I didn't word it as clearly as I should have, and I did gloss over your 3rd post where you emphasized you didn't care as much about reducing freewheeling diode losses, even though that is the only real benefit to a synchronous buck.

Yes, duty cycle does change from being relatively independent of load current in CCM to being directly proportional to such in DCM, but this is never an issue in the real world because the buck is the easiest switchmode converter to stabilize either way.

In contrast, during forced CCM operation of the buck choke (that is, maintaining current flow past the point the inductor current would otherwise go to zero), the synchronous low-side switch allows charge to get sucked out of the output capacitor only to shunt it to ground - basically, then, at near zero load current the upper switch is dumping charge into the capacitor only for the lower switch to suck it back out. So, not just pointless, but counterproductive*. This happens whether the lower switch is a MOSFET or IGBT, but with the latter type the one true benefit of synchronous operation - reduced voltage drop during the freewheeling period - is lost.

Furthermore, the benefit of reduced freewheeling losses becomes more marginal as the output voltage approaches the input voltage in the buck; generally speaking, one only bothers with the synchronous buck in low voltage, high step-down ratio applications like, say, supplying ~1V to a CPU from a 12V rail. Neither of these conditions apply to your application.



* - EDIT - actually, the converter can actually operate as a boost in deep fCCM mode, depending on how long the synchronous switch is on relative to the period, choke inductance, etc., but let's not complicate things further.
« Last Edit: August 22, 2019, 11:51:39 am by MagicSmoker »
 

Offline GKTopic starter

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: SMPS for dormant tube audio amp project
« Reply #16 on: August 22, 2019, 01:17:51 pm »
Will you guys please read my second post in this thread
I forced ccm with the low side igbt entirely for the control properties  ; constant pwm d/c at all load currents

Hmm, rereading my previous reply I see I didn't word it as clearly as I should have, and I did gloss over your 3rd post where you emphasized you didn't care as much about reducing freewheeling diode losses, even though that is the only real benefit to a synchronous buck.

Yes, duty cycle does change from being relatively independent of load current in CCM to being directly proportional to such in DCM, but this is never an issue in the real world because the buck is the easiest switchmode converter to stabilize either way.


The regulator needed to be stable without load. CCM fixes that problem. Having a duty cycle independent of load current also benefits the regulators transient regulation performance in response to sudden and large changes in load current - the control loop doesn't have to slew over a large change in duty cycle. I posed a whole lot of scope shots showing the regulators happiness going from zero load to 320W and then back again. It simply works. And the soft-start control as implemented with the TL494 overrides the control loop until the regulated output voltage is reached - that doesn't work properly in DCM. I can't remember every other design hurdle overcome to arrive at the final design completed almost a decade ago. I'm sure though that there are other ways of skinning a cat.
 
   
 
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: SMPS for dormant tube audio amp project
« Reply #17 on: August 22, 2019, 09:38:04 pm »
The regulator needed to be stable without load. CCM fixes that problem. Having a duty cycle independent of load current also benefits the regulators transient regulation performance in response to sudden and large changes in load current - the control loop doesn't have to slew over a large change in duty cycle. I posed a whole lot of scope shots showing the regulators happiness going from zero load to 320W and then back again. It simply works. And the soft-start control as implemented with the TL494 overrides the control loop until the regulated output voltage is reached - that doesn't work properly in DCM. I can't remember every other design hurdle overcome to arrive at the final design completed almost a decade ago. I'm sure though that there are other ways of skinning a cat.

Well, I can't argue with success, though there are, indeed, many ways to skin the proverbial cat (NB - I find it immensely funny the US and Oz share the same cliche). If you changed the compensation network to Type II (and did more than hand-wave the component values) you'd have no problems with transient response or other misbehavior around the transition from CCM to DCM, and the I haven't used the TL494 in decades - literally, almost 30 years - but there is a comprehensive app note on TI's website that covers how to use it: http://www.ti.com/lit/an/slva001e/slva001e.pdf.

I don't recall there being any problems with soft-start and DCM... then again, I don't recall much about the TL494 once I abandoned it and the SG3524 decades ago.


 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: SMPS for dormant tube audio amp project
« Reply #18 on: August 22, 2019, 11:02:47 pm »
Basically comes down to this: forced CCM is 100% the same control response over all operating conditions, period; whereas, when DCM occurs, the modulator gain at negative command (i.e., negative output current or falling Vout) drops to zero.  And so the dominant pole due to the filter cap shifts towards zero.  When it falls near the controller's dominant pole, they split and diverge (increased overshoot), and further still, they're pushed into the right half-plane (oscillation).

The choice then is: do you tune the controller for very low bandwidth, to keep stability, arbitrarily deep into DCM?  (You can't go on forever, because when load current goes to zero, the capacitor pole equals zero: an ideal integrator.)

Or do you tune it for CCM operation (prioritizing transient response), and allow the pole to shift into the right half-plane (i.e., it begins oscillating)?

Typical operation in this regime is, a short pulse (or burst of pulses), repeating at a frequency much lower than Fsw, and may be repeating erratically (i.e., the error amp's noise is effectively magnified, and shifted up, into switching noise).  The repeat rate being lower than Fsw is another way of saying the control loop has a RHP pole, that is about that far from zero.

Another aggravating factor is the minimum pulse width, which is usually limited by hysteresis in the PWM comparator, or driver, or of the current sense response, whether propagation delay of a peak current mode control, or bandwidth of an average current mode control.  In the latter case, of course, you're more likely to get a burst of pulses.

In a lot of applications, the increased ripple doesn't matter -- it's still well within nominal range, because the error amp is otherwise doing its job just fine, and the switching noise is adequately filtered either way.  Others, the consistent ripple from CCM may be easier to deal with (or filter) than the erratic DCM response, or the whine of the inductors or capacitors may be too objectionable.

CCM supplies still exhibit oscillation at very low output voltages -- minimum pulse width hits you regardless -- but the region out of the total SOA is very much smaller. :)

The downside is the constant "stirring" of reactive power, between input filter cap, and output filter choke.  The reactive power, divided by the effective Q factor of this loop, equals switching loss.

Which by the way, is a handy way to think of switching supplies, that I don't think is mentioned very often; the switching ripple is reactive power, and the total Q factor of everything it flows through, gives losses.  I used such an analysis here in regards to core Q: https://www.seventransistorlabs.com/Articles/Core_Loss.html and plotted for a few materials here: https://www.seventransistorlabs.com/Images/Powder_Core_Q.png The takeaway for lossy materials is, you avoid losses by using a small ripple fraction, which is to say using relatively few VARs for a given DC output.  That's how you can manage to use a lossy powder core that dissipates 1W, for say 10 VAR switching (Q = 10), and 100W of real output.

Alternately if we have very low-loss reactive parts, we can stir some of the device reactives (most often Coss/Cjo) into the switching reactance, and voila, we have a resonant power supply that can run at much higher Fsw, and even achieve higher efficiency, than a conventional (square wave) design would (for the same Fsw).

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf