simple PWM generator

I've used something similar in the past.

Yes, this is a well known circuit. I have used this topology since mid

1970s and have been posting PWM modulators using this on SED over the years (most recently for a lamp slow fade-up). If one can tolerate the frequency variation it is a wonderful low-cost solution. Wonder what is special about OPs?

piglet

I think I can now answer my own question. It is not the OP PWM generator that is special but instead it is the demodulator. JL has picked RLC values that give good ripple and phase shift.

piglet

Perhaps it's more relevant to ask how the simulation manages to generate a DC offset on the outputs, from the 2.5V ideal of the simulator.

- the source DC level doesn't 'measure' 2.500V, and that error follows through, with a gain of 20.

Maybe this is a 'real life' function of LTspice. . . simulating that nothing is perfect ?

You just don't expect to see errors in arbitrary sources.

RL

That's just a simulation demo, to see the recovered sine wave and the PWM distortion. It's a Bessel lowpass.

Actually I invented this PWM circuit (as I'm sure others have) ca

1970. Thinking it over, I probably should use something better, constant-frequency and linear almost to the rails. That will take more parts, but I may as well use a few more parts to optimize using a kilowatt power supply and a lot of mosfets and heat sinks and fans.

I'm going to build an open-loop class-D amplifier so I need good PWM. A closed-loop amp presents a negative impedance to the power supply, but open-loop looks positive. That could matter when things get hot.

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

On Thu, 05 Nov 2020 07:25:12 -0800, snipped-for-privacy@highlandsniptechnology.com wrote:

This is mostly brute force, but it is constant-frequency and will go arbitrarily close to the rails, depending on how RRO the opamp really is.

Version 4 SHEET 1 1500 724 WIRE -336 16 -368 16 WIRE -288 16 -336 16 WIRE 288 16 -192 16 WIRE 512 16 288 16 WIRE -368 48 -368 16 WIRE 288 64 288 16 WIRE -112 96 -128 96 WIRE -96 96 -112 96 WIRE -96 112 -96 96 WIRE 384 128 368 128 WIRE 400 128 384 128 WIRE -192 144 -192 16 WIRE -128 144 -192 144 WIRE 16 144 -48 144 WIRE 784 144 768 144 WIRE 800 144 784 144 WIRE -368 160 -368 128 WIRE 400 160 400 128 WIRE 800 160 800 144 WIRE -128 176 -192 176 WIRE 96 176 -48 176 WIRE 288 176 288 128 WIRE 288 176 176 176 WIRE 368 176 288 176 WIRE 512 192 512 16 WIRE 512 192 432 192 WIRE 640 192 512 192 WIRE 768 192 640 192 WIRE 928 192 848 192 WIRE 992 192 928 192 WIRE 1088 192 992 192 WIRE 1232 192 1168 192 WIRE 1376 192 1312 192 WIRE 1440 192 1376 192 WIRE 1472 192 1440 192 WIRE 368 208 288 208 WIRE -112 224 -112 208 WIRE -96 224 -96 208 WIRE -96 224 -112 224 WIRE -80 224 -80 208 WIRE -80 224 -96 224 WIRE 768 224 720 224 WIRE 880 224 848 224 WIRE 1376 240 1376 192 WIRE -112 256 -112 224 WIRE 16 256 16 144 WIRE 400 256 400 224 WIRE 784 272 784 256 WIRE 800 272 800 256 WIRE 800 272 784 272 WIRE 816 272 816 256 WIRE 816 272 800 272 WIRE 880 272 880 224 WIRE 992 272 880 272 WIRE 1008 272 992 272 WIRE 784 304 784 272 WIRE 1376 352 1376 304 WIRE -192 384 -192 176 WIRE 16 384 16 336 WIRE 16 384 -192 384 WIRE 96 384 16 384 WIRE 288 384 288 208 WIRE 288 384 176 384 WIRE 544 384 512 384 WIRE 592 384 544 384 WIRE 720 384 720 224 WIRE 720 384 672 384 WIRE 784 384 720 384 WIRE 928 384 928 192 WIRE 928 384 864 384 WIRE 288 416 288 384 WIRE 512 416 512 384 WIRE 288 528 288 496 WIRE 512 528 512 496 FLAG 784 304 0 FLAG 784 144 +5 FLAG -368 160 0 FLAG -336 16 +5 FLAG 512 528 0 FLAG 992 192 PWM+ FLAG 1376 352 0 FLAG 1440 192 DEMOD FLAG 992 272 PWM- FLAG 544 384 IN FLAG 640 192 TRI FLAG -112 256 0 FLAG -112 96 +5 FLAG 400 256 0 FLAG 384 128 +5 FLAG 288 528 0 SYMBOL Comparators\\LT1711 800 208 R0 WINDOW 0 52 -81 Left 2 WINDOW 3 26 -48 Left 2 SYMATTR InstName U1 SYMBOL voltage -368 32 R0 WINDOW 0 51 47 Left 2 WINDOW 3 59 79 Left 2 SYMATTR InstName V2 SYMATTR Value 5 SYMBOL voltage 512 400 R0 WINDOW 0 78 65 Left 2 WINDOW 3 36 101 Left 2 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V3 SYMATTR Value SINE(2.5 2.4 1K) SYMBOL res 880 368 R90 WINDOW 0 74 59 VBottom 2 WINDOW 3 84 57 VTop 2 SYMATTR InstName R2 SYMATTR Value 1G SYMBOL res 688 368 R90 WINDOW 0 -46 54 VBottom 2 WINDOW 3 -39 56 VTop 2 SYMATTR InstName R3 SYMATTR Value 1K SYMBOL res 1184 176 R90 WINDOW 0 77 55 VBottom 2 WINDOW 3 84 55 VTop 2 SYMATTR InstName R4 SYMATTR Value 10K SYMBOL cap 1360 240 R0 WINDOW 0 61 17 Left 2 WINDOW 3 57 51 Left 2 SYMATTR InstName C2 SYMATTR Value 2n SYMBOL ind 1216 208 R270 WINDOW 0 -40 64 VTop 2 WINDOW 3 -49 64 VBottom 2 SYMATTR InstName L1 SYMATTR Value 70m SYMBOL Comparators\\LT1711 -96 160 R0 WINDOW 0 68 -92 Left 2 WINDOW 3 40 -56 Left 2 SYMATTR InstName U2 SYMBOL Opamps\\UniversalOpamp2 400 192 R0 WINDOW 0 41 -38 Left 2 SYMATTR InstName U3 SYMATTR Value2 Avol=1Meg GBW=20Meg Slew=20Meg SYMBOL res 192 160 R90 WINDOW 0 75 53 VBottom 2 WINDOW 3 82 52 VTop 2 SYMATTR InstName R1 SYMATTR Value 5K SYMBOL cap 304 128 R180 WINDOW 0 73 57 Left 2 WINDOW 3 62 26 Left 2 SYMATTR InstName C1 SYMATTR Value 200p SYMBOL res 32 352 R180 WINDOW 0 57 66 Left 2 WINDOW 3 58 34 Left 2 SYMATTR InstName R5 SYMATTR Value 1K SYMBOL res 192 368 R90 WINDOW 0 72 56 VBottom 2 WINDOW 3 83 56 VTop 2 SYMATTR InstName R6 SYMATTR Value 20K SYMBOL voltage 288 400 R0 WINDOW 0 61 43 Left 2 WINDOW 3 59 79 Left 2 SYMATTR InstName V1 SYMATTR Value 2.5 TEXT 1072 472 Left 2 !.tran 2m uic TEXT 1040 360 Left 2 ;PWM Generator 2 TEXT 1056 408 Left 2 ;JL Nov 5 2020

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

You would have published it at the time then? Reminds me of other things which became standards such as baxandall tone control and tobey-dinsdale audio amplifier. Such things would likely be hidden behind patents now.

Certainly not. I design stuff to sell.

I noticed that but haven't investigated. Could be time delays or something. More likely comparator bias current, which can get weird.

I'm going to drive a transformer from a full bridge switcher, so I'll probably include some slow DC feedback to keep the differential output close to zero, so a little offset won't matter.

torsdag den 5. november 2020 kl. 16.25.22 UTC+1 skrev jla...@highlandsnipte chnology.com:

Technology, Inc

s

use an MCU with few ADCs and timers it can do everything you want plus all the nice "don't catch fire" features you can even get an MCU with the gate drivers and a few opamp/comparators in a single package

My box is all done, and I want to beef up the power amp boards, so going all analog is sensible. Doesn't need code!

The existing board has a current shunt and an isolated delta-sigma ADC that goes into our FPGA, so we could do some shutdowns or current limiting there.

The existing TI Class-D amp chip has too many fault sensing/shutdown features. We want to be able to really blast out power until the transformer smokes. Its thermal time constant seems to be about 30 minutes, so we could slam it for one.

Big fets! Dumb circuits!

fredag den 6. november 2020 kl. 00.05.52 UTC+1 skrev John Larkin:

ptechnology.com:

sch

and Technology, Inc

mid

he

the

t is

tor

all

s

if you already have an FPGA just use that, all you need is a couple of pins