Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Can Anyone guess the PWM ICs in this circuit?

Status
Not open for further replies.

marcbarker

New Member
Does anyone recognise what these ICs are ?

It's a PWM driver module for a DC motor speed control. It drives an IRFP460 MOSFET. All the numbers burnt off the capacitors when I dissolved away the epoxy resin potting compound.

You can see 3 ICs but I don't know what they are as the markings had been ground off by the manufacturer. The transistors hadn't though, they are mostly NPN, but there's a PNP and a 2N7000.

**broken link removed**

Here's some more info.
I decided earlier today to do some reverse-engineering, to find out what's inside an ECU. ECU is out of my treadmill excerise machine that stopped working.

There's a diode bottom left, but it's open circuit.

I have erased the PCB photo, but left the ICs, the transistors, polarised capacitors and an open circuit diode un-erased.
 
Looks like it's just a constant current controlled single PWM driver. You could redesign it from scratch using maybe 5 transistors, why go to all the trouble of reverse engineering a badly designed circuit that uses 3 chips 7 transistors??

You might want to post the PCB photos top and bottom, a lot of the tracks you have drawn don't make sense. Your +5v input is ciritcal to see where it connects to chips and transistors to determine pinouts but you have drawn it going nowhere.

The 16pin DIP is a TL494, a common PSU/PWM chip, pin7 is GND and pin 9 is a E output. You don't need much more than that 1 chip to build this entire circuit. Unless you're an idiot and want to use 3 chips and 7 transistors. :D
 
The 14 pin dip looks like it could be a quad OpAmp; LM324 type pin out.
The idea of pictures of both sides of the PCB is a good one. If you take the pictures straight on then it is a simple matter of mirroring the trace image, making it semi-transparent, and then layering it on top of the other in a photo editor.
 
Last edited:
well.... you have the makings of another Bob Pease......:D

the 8-pin dip looks like an LM311 comparator, the 14-pin is likely an LM324, and i agree that the 3rd chip is a TL494......

BTW.... Bob Pease is the engineer who designed the LM311, and is known for both his "hasty sketches" of schematics in his magazine articles, as well as his perfecting the art of "ugly construction", where he builds a circuit as a 3-D puzzle from the ground plane on up... his "ugly construction" technique is good up to and usually beyond 50Mhz....
 
Last edited:
actually what's more amazing is that he takes that and turns it into this:
 

Attachments

  • icmicrophoto.jpg
    icmicrophoto.jpg
    18.4 KB · Views: 1,190
Maybe someone has seen one of these before?

This is the whole Module, from where it came from.
**broken link removed**

This is the actual PCB out out of the module **broken link removed**

So far we've worked out that the middle IC is a 324 and the right hand one a TL494. **broken link removed**

Don't know what the left hand one is. It's not a 311 or single opamp.
 
are you sure there's 300V (or more) going into pin 2 of that chip and pin 5 of the 324?

maybe it's an optoisolated MOSFET driver?

a 2SK579 pinout is GDS, not DGS, and the 2SC9012 transistor pinouts are ECB, not EBC. and the PNP device should have a 2SA or 2SB number.
 
Last edited:
Yeah it's shocking, the 3pin device pinouts are way funky, the 5v line is still not shown properly (probably goes to that big cap?), the push-pull output pair don't even copnnect to the FET gate etc.

It's all over the place. If you can't even make a accurate diagram from the PCB it's gonna be a really hard job to reverse engineer.

Just design a new PWM circuit! It's the best option.
 
no... the big cap looks like it says "400V 330uf" and it's directly connected to the offline bridge...... that's at least 300V on that cap...

there are definitely some open nodes on the board, such as the +5V rail, and a couple of pads near the 8-pin. one of those open pads goes to a resistor...

you'd be better off making a wire list. once you make a wire list, it's easy to transfer it to a schematic. all of the 2 pin components get pin numbers of 1,2, with pin 1 being up for vertical components and left for horizontal components. for instance, starting at the top left, making the 20k resistor R1, the 24k resistr R2, and the 8-pin U1 a wire list would look like:

1.....R1.1_U1.7_R2.1
2....R1.2_D1.2_(empty pad)

and so on. each line on the list has all of the nodes a wire goes to. when you draw a schematic it makes it a lot easier to draw since you have a list of what is connected to what. i actually find it easier to start with the wirelists for the rail voltages, then the IC pins, then the transistor pins, and last, anything not yet listed. when working from a drawing or pictures, i use a highlighter pen to mark the wires i have done.

wire lists are also used by schematic and PCB layout CAD programs, and are also known as netlists
 
Last edited:
are you sure there's 300V (or more) going into pin 2 of that chip and pin 5 of the 324?

maybe it's an optoisolated MOSFET driver?

a 2SK579 pinout is GDS, not DGS, and the 2SC9012 transistor pinouts are ECB, not EBC. and the PNP device should have a 2SA or 2SB number.

Yes I'm sure the pin (via the SPD pot) goes to the + end of the (450V rating) capacitor. The pot measures 100K. Looks like a pot-down.

The 8-pin IC has a strange pin arrangement, it could be a driver I suppose good point.

I didn't see a 2SK579 on the PCB, where did that come from? Maybe I missed something. The Transistors are definately still marked with 'C9013 and C9012' and a 2N7000. I looked up a datasheet for them and it said it was 'EBC'. I am not absolutely sure I got the pin out for the 2N7000 right though.

The markings on 'japanese' transistors often omit the "2S"
 
Yeah it's shocking, the 3pin device pinouts are way funky, the 5v line is still not shown properly (probably goes to that big cap?), the push-pull output pair don't even copnnect to the FET gate etc.

It's all over the place. If you can't even make a accurate diagram from the PCB it's gonna be a really hard job to reverse engineer.

Just design a new PWM circuit! It's the best option.

I looked again at the "+5V" pin and it doesn't connect anywhere.

I looked again at the actual PCB closely and there doesn't seem to be any link between the PNP/NPN pair and the main Fet gate drive.

The drawing I've posted I'm sure is the same as the PCB. Luckily the chemical has not touched any of the metal, just some of the plastics.

This is getting difficult to reverse engineer yes you're right.

I guess there is some functions going on in the design that is the reason why there are so many parts in it. I'll guess that there is a speed Ramp control and a over-current protection too.

Might be easier to design another from scratch. Though I'm getting a slight deja-vu feeling, I'd done this kind of re-design thing before, after many hours eventually end up with almost the exact same circuit!

Maybe it can be done with a PIC somehow?
 
no... the big cap looks like it says "400V 330uf" and it's directly connected to the offline bridge...... that's at least 300V on that cap...

there are definitely some open nodes on the board, such as the +5V rail, and a couple of pads near the 8-pin. one of those open pads goes to a resistor...

you'd be better off making a wire list. <..>

The big cap is 450 V / 330 uF. The mains is full-wave rectified.

Wirelist (netlist) yes, I did start doing that, done a few nets, but keep getting interrupted these last few days! :)
 
sorry about that. i mistook the "579k" marking for a transistor marking.... didn't see the 2N7000 at the bottom of the pic.... also thought the C9013 was a japanese transistor number..... but since you have a c9012 which is pnp, then they must be house numbered, which means you may not find a data sheet for them. out of 4 data sheets for the 2SC9013, however, 2 of them have EBC, 2 of them are ECB, depending on the manufacturer....
 
out of 4 data sheets for the 2SC9013, however, 2 of them have EBC, 2 of them are ECB, depending on the manufacturer....

Ugh! dammit

I think these are ebc because the 0V and +12V rails seem to hit the right spots of the transistors. The pnp/npn pair and the 2 middle npn look that way.
 
It's not 2SC9013/9012, because 2SC are only NPN (2SA,2SB are PNP, 2SC,2SD are NPN).

I've seen BC series transistors marked with just Cxxx on the package. I just googled and found BC9102 and C9012 on the same page;
Datasheets 2008 5 13 20 | Datasheets archive

But really, how hard is it just to design this circuit from scratch???

It's only got one feedback (motor average DC current from a shunt resistor) so it either regulates motor PWM to fix a stable motor current, or it's just an unregulated DC PWM speed control with a over-current sense limit.

You could design either one of them in half an hour and probably have got the machine running long ago. :)
 
It's not 2SC9013/9012, because 2SC are only NPN (2SA,2SB are PNP, 2SC,2SD are NPN).

I've seen BC series transistors marked with just Cxxx on the package. I just googled and found BC9102 and C9012 on the same page;
Datasheets 2008 5 13 20 | Datasheets archive
Yes that is a very good point! I must admit that the source of this "2S" xxx had came from me in the first place, I was the one who incorrectly assumed there is an invisible 2S prefix. I still think the 9013's are NPN and the 9012 is a PNP. But I think this might be academic now because...

But really, how hard is it just to design this circuit from scratch???

It's only got one feedback (motor average DC current from a shunt resistor) so it either regulates motor PWM to fix a stable motor current, or it's just an unregulated DC PWM speed control with a over-current sense limit.

You could design either one of them in half an hour and probably have got the machine running long ago. :)

That seems the way to go then. I think a small PIC (with multiple ADC's and PWM) to generate the PWM. It does seem tempting though just to connect a 555 circuit with turny-knob! or even wire up a pulse generator into it with the earth removed! (no not really!).

In the original design there appears to be a soft-start and speed ramp control, can do that in the PIC too. This 'rate control' can be trimmed to taste with a potentiometer hanging off one of the ADC's.

The original circuit has a speed request input (representing I guess 0-100% speed = 0-5V). I am not sure how much PWM = 100 % speed, so I think there needs to be a "PWM span setting" to set the max. range of PWM for "100 %" speed request. I suppose that could be another potentiometer hooked into another ADC.

What I've done before for this kind of (entering empirical coefficients) thing with a digital design is to have a potentiometer with a button(s), to load the coefficient into EEPROM locations(s) pressing a button(s) or shorting pins together with screwdriver. The input pot. can of course be removed afterwards as is no longer required.

There's another thing needed is a motor over-current protection. On the original circuit there's a current sense resistor. I guess that when there's a VBE's-worth of voltage there, that's the signal to shut off the FET gate drive. I think this needs to be damned fast somehow, and if any software is involved it be interrupt-driven.
 
I would go for the 555 solution and just connect that to the speed pot. Definite "cool" factor.

As for the overcurrent deal, you could use one transistor that turns on when the shunt voltage reaches 0.6v and that drastically reduces the 555 PWM on period.

If you want slow "ramping" from speed to speed then I would go for a VCO instead, with a big cap on the V input and a trimpot between the speed pot and the cap. Easy as pie with a VCO but I don't know any easy way to do it with a 555.
 
I'm back on this job again.....
I would go for the 555 solution and just connect that to the speed pot. Definite "cool" factor.

As for the overcurrent deal, you could use one transistor that turns on when the shunt voltage reaches 0.6v and that drastically reduces the 555 PWM on period.

If you want slow "ramping" from speed to speed then I would go for a VCO instead, with a big cap on the V input and a trimpot between the speed pot and the cap. Easy as pie with a VCO but I don't know any easy way to do it with a 555.

I did think of a knob at first, but I can't bring myself to drill a hole, plus I'd like to sell the machine later in year.

Hmmm... I think you mean a 4046 with a capacitor hanging off the VCO control pin (555 with fixed pulse length) ? component count might be going up

I'm drifting to a PIC single chip solution. If I do, maybe I'll have to add a 555 as a driver anyway, as a bistable, drving the fet, with it being reset if the current sense exceeds a certain level (sounds like going in the direction of a standard current mode smps!)
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top