 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Electronics : Variable input voltage bat charger
  Page 7 of 8   |
|||||
| Author | Message | ||||
oztules Guru  Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
With pins 15 and 16. With pin 16 grounded, and 15 tied to vref, the op amp is wide open.. (it's output is 0v). or put another way.. When the inverted input is higher than the non-inverted input, the result should be low on the output... this means max pulse width all the time. This is OR'd with the output of op amp pins 1 and 2. (remember pin3 low = max pulse width. Pin 3 high means min pulse width... that triangle/voltage thing again) If you reverse the pins 15 and 16, the result will be high.... it will shut down the PWM regardless of the pin 1 and 2 results. The highest voltage wins (regulates the pwm downwards) If you make sure that v- is not earthed, you can see all the pulse waves on all the pins on the scope (on the secondary side only without isolation). Check to see if you have stable (no noise) on your pin4. It is a long way above ground, and may be prone to noise (long shot but easy to check) EDIT: and pin 12 vcc Gee, now I think of it....... What voltage is on vcc on the 494. Is it possible your feeding some of the 48v into it? The chip will shut down and start up again....... If it is only fed from the flyback, then what is it's voltage? If this is a conversion, (not your own board ), you may still have the 48v rail driving into the 494 via the original circuit.... check the 494 vcc with the scope.. is this the noise etc etc... The problem will be simple and staring us in the face, but without being there, it takes time to follow hunches... keep on keeping on  We'll find it sooner or later...
I think the over current circuit is the first thing that I try to do... stops the fire works. .. but we need to find your instability first I guess. 
.........oztules Village idiot...or... just another hack out of his depth |
||||
sPuDd Senior Member  Joined: 10/07/2007 Location: AustraliaPosts: 251 |
All pins are clean when running 100%. Output is clean when 100%. Noise is a result of failure to PWM. 12V to 494 comes from the original aux PSU. At 100% duty its stable & clean. 48V OP only shares a ground for reference. No point current limiting if it won't voltage regulate. I must be doing something wrong - it bugs me that I can't see it. nickers It should work ...in theory |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
Ok Spud.... no more Mr nice guy. 1. Grab a 100uf capacitor and dump it across pin 2 and pin 3 (+ to pin3).. (yes we will bypass the feedback components.) 2. 4.7uf to ground from pin2 3. Stand back and switch on. If it is in the chips power rails, or feedback, this should knock the noise out of it. I had a play with one of mine, and tried all sorts of ways to induce instability... it is not hard. However, with the 100uf in place and the 4uf from 2 to ground... it was just impossible to get it to behave badly... and I learn't a few things along the way. We don't need to make the feedback loop very sensitive at all. In fact if you short pins 2 and 3, you should get a pwm train less than full bore.. and no instability (or real control for that matter). If all of that doesn't work, then you must have a massive earth loop somewhere. This can only be problematic when the op-amps are active, and the difference between 1 and 2 is less than the noise.. so the noise takes over.... hopefully not. Anyway, see if that gets control... and I may use those values from now on, makes a very stable device. I can touch all or any pin with a probe, and there is no flutter at all..... impressive. Now this one runs from 1.7v to 29v no problems, and I have seen 25A at 10 or so volts as well (as that big vac motor speeds up from stop to 15000 rpm) Then the current sense... It is pretty low impedance anyway, and it too benefits from changing the feedback loop from r/c to big c. I will change that other diagram when I test some more and see there are no major draw backs. You do have some kind of load on it??? a must. .........oztules Village idiot...or... just another hack out of his depth |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
Hmmm, been pondering it all arvo. I can only think that its noise that starts at less than 100% duty (reason to complex) and causes it to spit the dummy. I'll nail the inputs down with caps, and drop the gain to almost nil. I have nagging sense of dejavu. I have a nichrome heating load that I can adjust from 60ohms to zero. It always has a load. I never run any psu without a load. sPuDd.. It should work ...in theory |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
OK, I added the caps like you said. Got slow response, but no change. Added 10K from pin2 to pin3 to reduce the gain. Helped slightly. Refitted the original transformer and diode pack to make a 13.8V output. No better compared to the 48V one I wound. Also fitted everything to the PCB, big caps, trim pot, filters etc. No difference. Found a lot of the problem was C22 was missing. Fitted a 10nF as that was the original. I can now trigger the CRO off the waveform, but its got false triggering on each opposite half. Funny thing is, it now runs very hot on the transistor heatsink. Its too hot to touch within 15 seconds, with 13.8V output across 4ohms. The nichrome load I was using was causing some ringing due to the inductive nature of the wire windings. Replaced C9, C10, C11, D5, D6, Q3, Q4, and the TL494. No effect. CRO on both outputs from the TL494. Notice the false triggers and severe ringing on one half. No amount of stuffing about will get rid of it. 
One turn of wire around the transformer shows this. It explains the transistor heating. It might as well be a sine wave. 
Pics of the PCB: 
My home made flyback PSU is cleaner and more reliable than this, and it doesn't even have any voltage feedback from the output to the IC!! Let alone any secondary current limiting! Overall, I suspect this is a cheap design, which relied on the 120mm fan that was in the case to hold it together. The original design had no current limiting. I'd be better off building my own from scratch. I'm going to add 1K pull downs to the TL494 output, as I suspect that is where they saved another 2c. sPuDd.. (Gizmo - the pics all had the same specs, uploading to the server resized all but the first - weird) It should work ...in theory |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
Good grief.. those wave forms are a shambles. Yes the big caps on the feeedback kills the response time, but should have quelled any oscillations in the feedback section, and allowed fault tracing. Don't use it in full time use, as the current response will be too slow to save you. The voltage is not that critical, and a .25 second re-adjustment is usually inconsequential. I don't know how to help now... it should have been fine. I can't understand how the original is running with that wave form. Something is seriously simply wrong. That circuit has no pull downs, but the one I am playing with at the moment has 1k5 pull down and 3k9 pullups on the 494. Is your mode pin floating? (13).... C11 hasn't dried out has it??...It just don't make sense
..........oztules Village idiot...or... just another hack out of his depth |
||||
| Dinges Senior Member Joined: 04/01/2008 Location: AlbaniaPosts: 510 |
Spudd, Have you tried replacing D5 or D6 (depending on in which channel you have the ringing problem)? If that doesn't help (don't just measure the diodes with the diodetest function of the DMM; replace), try replacing the other diodes in that area (D7, D8, D14, D15) and C11. May want to try replacing Q3 and Q4 too, if the above measures don't help. Out of curiosity... if you remove Q3 and/or Q4 (the transistor which is in the ringing channel) and you then scope the output of the TL494, I assume that the outputs (pin 8 & 11) are then perfectly clean? Peter. |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
Oztules, the unit had board spots for pull downs. Fitted 1k and got an angry bumble bee. Will change the pull ups to 3.9K (from 2.7k). The function pin is tied to Vref. I'd be surprised if a green cap dried out, but hey - this is YumCha. Dinges: Replaced C9, C10, C11, D5, D6, Q3, Q4, and the TL494. No effect. I'm of the opinion that I'm chasing a crap design rather than a fault. But the variation in drive channels tells me almost otherwise. Personally I think I scrap it and build my own. At least it'll work. sPuDd.. It should work ...in theory |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
Sorry Spud, I thought c11 would be electro. (1uf 50v) If you give up with it... send it to me. I'm dying to know what is going on with it. It's just making no sense. ................oztules Village idiot...or... just another hack out of his depth |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
Oztules, I'd send it to you, but it irks me so much I can't put it down. Bad habit. I ditch it when it gets to me, but I just keep going back for more. I'm going to begin shotgun repairs, concentrating around the isolated drive and main transistors. I suspect its a drive issue. I will power up the aux 12V line on its own and see if the TL494 can drive clean or not. Heck, might even learn something in the process. sPuDd.. It should work ...in theory |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
Replaced all the diodes on the HV side, all the remaining diodes on the LV side. Checked all resistors. Removed the main transistors & checked. I powered the output from a bench supply, so I could make it run and scope the board. When set it to 13.85V the IC outputs a textbook clean 50% duty (25% from each half) which transpires to a nice clean drive signal all the way to the base of the output transistors. I did notice only 1Vp-p drive to the bases, but I suspect this is OK, considering they only need 0.6V to turn on. Am I right? The thing turns to sh*t once powered from itself. I've filtered to the wazoo, changed all the caps in the aux PSU, nada. The switching noise that I suspect is causing the problems appears on the output, feedback, aux and Vref rails to the same value! About 1Vp-p, regardless of anything. Its almost as if its being induced through the PCB itself. Top trace is the output of one channel. Bottom trace is from the centre of the isolation drive transformer. Both are while power from a bench PSU. 
I'll have to put this piece of junk aside and try another one. sPuDd.. It should work ...in theory |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
At least you have isolated the problem.... Perhaps track down how the aux gets it's neg grounded??. Village idiot...or... just another hack out of his depth |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
Earths, yes. Main negative out is heavy tracked (with extra solder) right through the spine of the PCB to the aux negative. I also added a closer ground connection to the track that runs around the pcb edge (removed the other one) so the run was shorter and more 'star' like, less like a loop. Thought I'd prove it was a design issue. Pulled out another unit same brand & slightly larger. One wrap of wire around the main core (very good trick Oztules) fire it up. Bugger. It works fine. CURSE YOU YAHA!! 
It runs the same load for ever while the transistor sink gets barely warm. No fan. OK, I'll mod the new victim and see what happens. At least I've proven it works 100% before I use it. I vote we name the core wrap test the “OzWrap” ? TulesTest? 
It should work ...in theory |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
Hehehe... should I wait by the mail box???
..........oztules Village idiot...or... just another hack out of his depth |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
Yes, there will be two of them. Did a total mod on another WORKING unit. Same thing. So there must either be some trickery involved in the previous design to make it run stable, or there is something wrong with our design. PS: I'll trade you the PSU’s for the hours of my life I'll never get back 
sPuDd.. It should work ...in theory |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
Keen to get to the bottom of this Spud. If your serious, send it to my name (from profile) c/- Whitemark Post office, Flinders Island Tasmania 7255.. that will find me. If I can get it to be sensible, you can make the mods on the second one.
...........oztules Village idiot...or... just another hack out of his depth |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
I woke at 3am and started playing with it. Not a good sign. I think I might send a few PCB's to you, modded and working un-modded. Hell I got plenty of them! sPuDd.. It should work ...in theory |
||||
| sPuDd Senior Member Joined: 10/07/2007 Location: AustraliaPosts: 251 |
Oztules, box is on its way to you. Australia Post. ~300mm cubed. 2x modded 13.8V PCBs & one very similar working unmodified unit for comparison. Also some North Queensland treats 
sPuDd.. It should work ...in theory |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
Ok, will look out for it soon.
...............oztules Village idiot...or... just another hack out of his depth |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
Well Spud, as you may be aware, I only get to "town" once a fortnight.... and I don't even relish that.... However, I have it on good authority that there is a parcel there fitting your description. A friend has picked it up and will bring it up today..... so we should know something by tonight hopefully. I was not due to hit town till Wednesday (ship comes in on Monday/Tuesday, so there is food on the shelves for Wed/Thurs...hence every second Wednesday is town day.) Will report later today hopefully. .......oztules Edit.... this thread continued here Village idiot...or... just another hack out of his depth |
||||
| Page 7 of 8 |
|||||
 Print this page |
