#1232

Episode 1232 chip of the day
let's look at this strange chip
Be a Patron: / imsaiguy

[@d614gakadoug9]

It is a pretty straight-forward IC but it is also a VERY old design, dating from the days when bipolar transistors were used as the power switches. "T" is for toggle, not trigger. The waveform was perfectly synchronous, but the scope trigger was set to the toggle input, not the FF output. Transconductance amps are moderately common in SMPS controllers. They simplify matters considerably with a circuit such as this because you can connect multiple amps together at their outputs and pull the output down with impunity for hold-off or soft-start. Frequency compensation is done with components connected to ground, rather than in a feedback network. It was doing pulse width modulation just fine with the single collector.The maximum duty cycle on each output is limited to less than 50% (well, Duh!) The intent is that it would be used in a half-bridge, full-bridge or push-pull converter. No, it is not way over complicated. It is quite versatile and clearly beyond the thought processes of someone who has never actually done switcher design beyond simple single-ended stuff. There are certainly better controllers around now for high-performance DC-DC conversion, but they are even more complex internally than this part.

[@W1RMD]

Thanks! I love your "Chip of the day" videos!

[@byronwatkins2565]

The most efficient, resonance based, zero-crossing converters can also be excited and controlled by such a more generally applicable chip. These output transistors can also supply gate signals to hi-power MOSFETs and IGBTs for more applications. Pulled up to 5 V, the collectors are also suitable for H-bridge inputs. The comparator flexibility is also suitable for power factor correction. But usage does require more skill and more circuitry.

[@jj74qformerlyjailbreak3]

I’m like you. I like printing out DataSheets. It’s way faster. I can only imagine your file cabinets. Always great content. Definitely a channel I have on my android box. God Bless

[@SimoWill75]

Absolutely! For my projects I still print, punch and ring bind. They are full of hand written notes on page borders(reminders/warnings etc), post-it notes, page tabs, highlights in different colours etc etc. Some old-school methods are still way more time efficient.

[@theoldbigmoose]

I've pondered that data sheet for hours... your video sure helped!

[@jainmanish123]

Thanks for the video. Interesting demo of the functionality.This is also used in dc to ac inverter designs where the two outputs are needed for the AC waveform kind of push pull, that was my first encounter with this chip.

[@andymouse]

Yeah very bizarre ! never seen it before and thanks for the walkthrough....cheers. (love chip of the day !)

[@Edisson.]

Hi, these videos are absolutely amazing, on the one hand, I learn a lot of interesting information about a specific type of circuit, and a practical example of wiring replaces several professional publications. It is extremely useful that it is someone else's point of view, we all think a little differently about circuits and what one cannot see, another can see, when these thought processes are connected, a complete package of information about the given circuit is created. I don't know if I wrote it exactly as I mean it, I don't know English very well and I'm helping myself with the help of a translator. Thanks for these type of videos and have a nice day 🙂Tom

[@jp040759]

Looks like the TL494 chip. Cute approach to discovering how this chip works.

[@SkyhawkSteve]

These used to be the building block for a wide variety of SMPS. IIRC, the two complementary transistor outputs were used to drive the push-pull primary of the isolation transformer. The T.I. datasheet for the SG3524 shows all of these configuration options.

[@argcargv]

The architecture of the chip allows it to be used in resonant converter type push-pull configurations. You can connect the two transistors a totem pole arrangement and the output can push and pull current without concern for shoot-through. This type of configuration is good for resonant converters that have lower noise than flyback style converters. It is odd that the datasheet didn't show an application that takes advantage of this. I looked and the only application note I could find that sort of exploited this is the application note 288. But there seems to be very little on the interwebs on this obvious use-case for this chip.

[@analoghardwaretops3976]

They definitely don't have shoot through issues..but equally dangerous is power device failure ( dead short..that kills the other one too) due to possibility of one sidedflux buildup ,leading to core saturation in either normal push pull pwm or even when used in resonant mode...because of absence of " Volt-Sec. " balance ( that is a must.) ...either internal to chip or added externally.

[@ke9tv]

It's a great chip for its age. Lots of switching power supplies use it. Usually, you use an RC network on pin 9 to stabilize the loop filter. The OTA on pins 1/2 is for voltage sense, as the voltage gets high, it pulls down the pulse width. The one on pins 4/5 is for current sense, you measure the drop across a shunt. It gives you foldback current limiting, not just a hard limit. The separate collectors are for push-pull driving a transformer primary (or driving the gates of MOSFETs or IGBTs that in turn drive the transformer). Often you see them pulled up to drive an H bridge. Normally, you put an RC network on pin 9 to stabilize the loop filter. Oftentimes, your voltage and current feedback come back through an optoisolator, and this chip is running on the high side. of the transformer. The reason for the complexity is that the chip tries to do everything. Nowadays you buy a chip for the specific circuit topology you're using.

[@zyeborm]

Heh 20 odd years ago when I was in high school just starting to look at SMPS I designed and breadboarded a circuit that looked almost the same as this chip. Only difference I had was not using a ramp generator. I was only doing buck conversion so I just used a flip flop and a clock with a comparator to drive it all. It worked surprisingly well really.

[@roboanalogtom]

I've been playing with single op amp PID circuits and would love to see you analyze them. The PID output feeds a single op amp PWM (square wave generator with an input to the non-inverting side instead of a normal voltage divider) for a complete lm358 PID/PWM controller for heat/cooling control. The PID circuit is from "Realization of a Low-Cost Op-Amp Based PID Controller" by Ibekwe (2018). edit- the input I'm using is a 10K thermistor/ 4.7K resistor as a voltage divider and the output is driving a 2n2222 for a fan or heating element.

[@M0UAW_IO83]

They've gotta be 30+ years old at least, they were used in some variants of the original IBM PC PSU and they were old hat then... Good old days though, when chips were designed to be flexible and offered the designer options on configuration rather than app notes with dire warnings about deviating from ref designs.

[@sebastian19745]

So, if you input some AF signal at pin 9, with this IC can make a class D amplifier? Maybe driven a H bridge? Also, it can be used as a (variable frequency) pulse generator up to few hundreds KHz with variable duty cicle. Nice.

[@BrananPR]

The semi-random phase of the flip-flop output is almost certainly due to the trigger not consistently hitting an "on" or "off" pulse when triggering on the oscillator. You could probably get it to stabilize by playing with the hold-off

[@Manf-ft6zk]

I was waiting for it, enjoying the real bottom up explanation of the chip. The proper way was given here: "Lets go and trigger on channel 2" kzbin.info/www/bejne/qGm4pHaggbRnhZo.

[@absurdengineering]

Simplest just to trigger from CH2. CH1 ramps triggers on either 0 or 180 degrees of the output waveform and thus can give either polarity on CH2. Hold off would cure it too, just more finicky.

[@t1d100]

Might it make interesting beats for the new synth?

[@MikaelIsaksson]

Hmm, if you can manipulate some dead time between the flip flops you could use it for an H-Bridge.

[@analoghardwaretops3976]

3524(200kHz)/ 3524D(500kHz),. both versions have max.duty cycle of around (94%....47+47) ...which is sufficient dead time (6%...3+3 ) even for slow bipolar transistors that were available in 1980's when the chip was introduced...to increase this dead time some more suitable xxx nF cap connected to pin ( osc. out ) can be selected as per datasheet graph that shows plot of cap value vs dead time in uSecs...this will bring down max duty cycle further... another way is with voltage clamp on Compensation pin......but for H bridge seperate high side & low side dead time has to be designed seperately with external circuitry...

[@johnyork2366]

I see further down in the comments alotta people seem to be confused as to what this would be used for.When I watched the video it hit me right away the applications for this. Now...that being said I've only started really learning about electronics about 2 years ago and maybe off the mark fully...But as to the set-up-wouldnt a massive app for this be a AC/DC-HF-PWM welding machine?specifically GTAW & SMAW freq-balance and foreground/background Slope control when pulse welding with HF??? now like I said:I'm very new to electronics!lol I could be completely off the mark...but the functionality of this circuit as an integrated one into a hi-amp welding set-up would be EXTREMELY precise in the welding of nonferrous metals-namely ESPECIALLY Aluminum! :) Are any of you guys welders? Like I said:pay no mind if I seem the a kid trying to talk to adults at an adult party!lol If I'm wrong,tell me.Either way awesome set-up!Excellent video sir!

[@analoghardwaretops3976]

Certainly for welding app.control circuitry..but with lots of limitations & low reliable performance, if used as a standalone ..... For high current applications , current & temperature control in the feedback loop has to be designed with additional external components / circuitry ...to reliably work ..

[@frankowalker4662]

Heres an odd question for you. Whats the lowest frequency of the square wave while remaining stable ? (say around .01%)

[@IMSAIGuy]

no idea, this is not the circuit to use if you want stability

[@frankowalker4662]

@@IMSAIGuy Cheers. Just wondering. :)

[@d614gakadoug9]

0.01% frequency stability puts you into the realm of a quartz crystal. Even ceramic resonators aren't generally that good and their initial accuracy is typically quite a lot worse than that. If you need something low frequency you can simply use counters to divide your crystal oscillator frequency. If you need something adjustable you've go a problem. I would say the "best" solution there these days is to use a direct digital synthesis IC. They don't inherently produce square waves but you can use a comparator on the filtered output. I think you can find complete modules that will do that fairly easily. If you need a super-precise clock signal at reasonable cost, a "GPS-disciplined" voltage-controlled crystal oscillator is not very complex and quite cheap. You start with a VCXO that has a narrow tuning range and use it in a phase-locked loop that compares the signal from a GPS receiver module (I'd recommend something based on a u-blox product) with the output of the VCXO and phase locks the VCXO to the GPS signal. You can program the clock output from a u-blox module over a fairly wide range (I don't recall, but detailed in datasheets). That might be usable directly for what you want. Of course you do have to be able to receive from the GPS satellites.

[@frankowalker4662]

@@d614gakadoug9 Thank you. 👍

[@teslakovalaborator]

Looks like a SG3524 just with a different name.

[@analoghardwaretops3976]

SG ...Silicon General...they too have the SG 3524D version ..this " 1st. gen ..Voltage mode PWM " controller was manufactured by many companies but all had to carry the same generic " number name"...

[@boonedockjourneyman7979]

Why was this thing developed? There would have been a major market to justify something this unusual.

[@absurdengineering]

It was a common SMPS chip. I still use it sometimes for demo or one-off things. Got a tube still full of them :)

[@analoghardwaretops3976]

Everywhere there's always a beginning...& everything is new even to the " experts " who developed them...nobody can foresee any flaws/ limitations.....hence we have..." new & improved...." / " version 1.1.xxx/ version 1.2.xx etc

[@d614gakadoug9]

It was a perfectly good controller suitable for a broad range switchers. It seems too many people these days have never seen anything but simple buck, boost and flyback converters. This could be used for any of those but also for push-pull, half-bridge and full-bridge circuits from a few watts to kilowatts. There are better ICs around now and many of them are substantially more complex internally than this one.

[@stevenbliss989]

Looks like a better TL494 to me! :)

[@stevenbliss989]

OK, the TL494 is 5% and the LM3524 is 1% ref accuarcy. Price wise TL494 is 1/10th. the price though! :)

[@analoghardwaretops3976]

Internal logic of ..494/594 has multiple pulse suprsession ( in one switch phase ) whereas the.... x524 does not... because the ..x524 comparator o/p directly enables/ disables the o/p ..but this & other limitations are overcome in the ..x524D ..this makes the .... 24D an even more reliable chip than the ..494...

[@stevenbliss989]

@@analoghardwaretops3976 Not aware (until now) the difference, but I would have though multiple pulse suppression is a good thing? BTW, not a fan of single tranny output controllers like the 494, has it's place, but for a now mosfet world, they are a pain. UCC3825 and SG3525 (the poor cousin when price really matters) are good to me. Any other types you've played with that I could look at? ....there are SO MANY controllers out there it's INSANE! :)

[@analoghardwaretops3976]

@@stevenbliss989 multiple pulse suppression is a must and is implemented in all later day designs . The multiple pulse suppression is achieved because the pwm comp. out is now fed to 1 input of the R-S ff in the x524D version ,which is not the case with the 1st. gen. x524...

[@stevenbliss989]

@@analoghardwaretops3976 I thought that might be the case, thanks :) BTW the TL494 is crazy cheap, but the SG3525 is not much more, and for me vastly better. But I still love to use the UC3825 when $ don't matter, ...high gate drive, proper current control ability (the SG35325 sort of has it, but I do not trust it) ..but $$$ Any other PWM chips you use? Always looking for others experience with chips I have not played with.

[@AnalogDude_]

a TL431 on steroids?

[@paulcohen1555]

No resemblance between them.

[@AnalogDude_]

@@paulcohen1555 there is a comparator and internal reference of 2.5 volt inside the TL431, check the video from GreatScott on it. he shows it actuating.

[@analoghardwaretops3976]

TL431 is a 3 pin high accuracy precision reference regulator that's also temperature compensated... it's not a pwm controller... the ...24/...24D have an internal 5 V ref. that's not so tight in specs...and any external voltage divider adds it's own tolerance errors...to make it a far less reliable reference .

[@AnalogDude_]

@@analoghardwaretops3976 i said on steroids : )

[@d614gakadoug9]

@@AnalogDude_ There is an amplifier, not a comparator, in the TL431. It _can_ be used as a comparator but it was designed as a linear amplifier with pretty decent characteristics. Clues: it calls it an amplifier in the data sheet and there is a frequency compensation capacitor in the schematic of the amp.

[@theoldbigmoose]

I've pondered that data sheet for hours... your video sure helped!