1
\$\begingroup\$

I'm venturing into building switching PSUs. Voltages on the order of 5-20v, current 1-10A.

To be optimised would be the efficiency of the power MOSFET(s), hence Vgs and Vds rise and fall times.

Ive got a 20MHz 48Ms/s Hantek USB scope - the cheapest I think. An ancient 500kHz Jupiter function generator. And FWIW, plenty to build quick circuits out of.

How can I do this cheaply? The oscilloscope would be great I think, if it weren't for its inability to zoom in on edges! But useless if switching only represents 2% of the time? Might it be suggested I invest in other equipment? Or is there a clever little an analog / hybrid digital circuit I could bread-board up? (straight forward, effective and cheap - optimistic?)

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Measure the temperature of the MOSFET as you're slowly increasing the switching frequency. When the MOSFET starts heating up quickly, it means that the switching losses are becoming significant. Accurate: Hardly. Straight forward, effective and cheap? Arguably yes. \$\endgroup\$ – Dampmaskin Apr 22 '18 at 10:50
2
\$\begingroup\$

Given the paucity of your equipment you may be better off living mostly in simulation world (eg. LTspice, with the most accurate models you can find) and attempting to verify operation by indirect measurements when you actually build something.

Since switching power supplies (for static loads anyway) involve periodic signals you could consider building a sampler (here is a link to an article), an ancient technique that works well for such signals, however personally I'd be trolling for an older working analog 200MHz oscilloscope.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ That key: An analog scope for this sort of thing is superior in most ways. \$\endgroup\$ – Peter Smith Apr 22 '18 at 15:37
0
\$\begingroup\$

I used the AM685 latched-comparator to build a tracking ADC with 0.4 nanosecond aperature time, 40 years ago, so I could measure waveform flatness to a precision better than the TEK equipment (real time and sampled) of that era.

You will have to provide the delayed clocking waveform.

I suggest you buy the scope.

schematic

simulate this circuit – Schematic created using CircuitLab

WIth the AM685 output simply being LowPassFiltered, with no amplification nor level-shifting, the input signal from 75 Ohms must be within the range defined by logic0 and logic1 output voltage levels.

Linear Technology has the LT685 ecl comparator, with schematic if you want. The datasheet does not state the aperture time (the on-to-off time). The minimum TRACK time pulse-width is about 3nS.

The track-hold behavior occurs (see the schematic) right at the input diffpair collectors; thus some transient charges will couple thru Cob to the analog input pins; keep the impedances low, so these transient charges are quickly discharged.

\$\endgroup\$
2
  • \$\begingroup\$ That's nuts! I'd love to know how you did that. I don't suppose you've blogged about that? \$\endgroup\$ – CL22 Apr 24 '18 at 9:37
  • \$\begingroup\$ We want the ECL output of the AM685 (or modern equivalent) to drive a DC_feedback loop to one input pin of the AM685. The fast signal, to be characterized, enters the other pin; system noise and thermal noise will cause HUNTING around the "equality" voltage, and we'll get dithering. The key is the external timing pulses. This was for an FM-transmitter, spaceborne, and I had to show a flatness of 1% given 5nS edges at 100MHz baud. Basically you generate some symbols, perhaps 1010101 pattern of bits, and walk the HOLD edge along the bit stream. The feedback is DC, and a DVM is the readout tool. \$\endgroup\$ – analogsystemsrf May 5 '18 at 4:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.