Resonant Converter Control - UC2875

I prefer the more modern ucc3895 chip.

--
 Thanks,
    - Win

Scott means phase-shift PWM, not to be confused with phase-control ac-line power control, as in the old triac days. In case that's what you were thinking of, dna.

--
 Thanks,
    - Win

Hi! I am building a resonant power converter for my degree project using a bunch of IGBT's in an H-bridge and require them to be phase controlled. This chip does the job, but i need a had wiring the thing together! any help would be greatly appreciated

many thanks scott :)

snipped-for-privacy@ntlworld.com

bunch > of IGBT's in an H-bridge and require them to be phase controlled.

No you're not.

Give me the name and address of the miscreant that gave this as a final year project and me and my mates from 'Leather Dyke Riding Bears R Us' will come round and transform his ass.

DNA

Oh shit, I have to admit that black women have tempting buttocks. It's not a substitute for love and that sort of stuff but they are immensely warm and cuddly.

I'm just guessing.

DNA

a

controlled.

final

a

Whoops, Semi-Sober mode on. Yup, since UC2875 is in the subject line then Scott does mean phase-shift PWM. I realised that.

I just thought that such a converter for a final year degree project would be a major challenge and would question the sensibility of the person setting the project. The mention of IGBT's makes me worry a bit more.

Now, you've built these things so if you asked one of your students to take one on as a final year project it wouldn't be so worrysome. You know how to do it and you know what's needed to get the job done.

Your student wouldn't be asking questions in a newsgroup.

However, Scott has asked the question in a newsgroup. OK, 'good' use of available resources but, if he needs advice, his first port of call should be the person who set the task.

Unless that person, for whatever reason, isn't up to the job.

DNA

It probably looked great on the simulator ;)

Sorry - couldn't resist.

Bob

Reference me in you thesis then:

Firstly, ensure that the DC bus inductance is nice and low. IGBTs DO NOT LIKE VOLTAGE OVERSHOOT and will die. To whit, L*dI/dt is the killer. If your big DC bus cap is 6" away from the H-bridge and connected with a pair of untwisted wires, expect many dramatic failures. Current flows in loops, keep them small (inductance is proportional to loop area). You cant do anything about the internal packaging inductance (other than buy a package with lower L) and that sets a limit on Vdc wrt Iout.

Then watch what happens under fault conditions - IGBTs limit fault current (like FETs) due to Vg, but often its 10x rated current. So dI gets a whole lot bigger when the IGBT turns off under fault conditions (say by a desat detector), and LdI/dt then blows the IGBT up. So L must be minimised, and using a large Rgoff for faults is helpful - dt gets larger.

The gate wiring itself needs very low inductance - Mr Miller tries to turn IGBTs off when you turn them on, and on when you turn them off (look at slope of Vce). If you are stupid enough to use a very large Rg, you will discover this in the first dozen or so switching cycles. Normally its Lg that causes problems here. Over-current fault turn-off can actually yank gates high enough to break them.

Its not so much what you do, as how you do it :)

Cheers Terry

I did a 3-month stint at a major university in NZs largest city, it was edifying. The EE dept had a set of design rules for PCB layout, that stressed using single-sided artwork (so they could make it themselves) and widely spaced components. My job was to troubleshoot a 3-phase phase-shifted IGBT dc-dc converter, that had an MTBF of about 3 minutes.

The problems were caused mostly by poor construction (wrt layout) but the PCBs were atrocious. I especially liked the 12V:24V single ended

100kHz dc-dc converters that used a 1kV FET. The original design came from the company I used to work for (45V FET) and used Cds to resonantly reset the xfmr. Turns out that 1kV FETs behave differently. And the output electrolytics had about 30 ohms ESR. not good. So I re-did the PCB layouts as DSPTH, spec'd appropriate components and removed huge amounts of wiring inductance, and the design worked first time. still does.

I was aghast that NOBODY seemed to know ANYTHING about real component behaviour.

Cheers Terry

Hi Bob,

thats a very good point - it certainly did. It also looked good on the schematic - I did not make any changes to that, other than part numbers (OSCON caps are great BTW). All the real work went into the layout/construction.

Cheers Terry

Hmmm... I hadn't heard of OSCON Caps so I looked here:

We have been using Taiyo-Yuden MLCC's for our low ESR requirements.

Bob