DFGT Circuit board help

  • Thread starter Thread starter luizsilveira
  • 71 comments
  • 26,358 views
Sick Lenny
Just got my new TVS diode(s) yesterday. Question is how do I tell which way to install? Which side goes to V+ and which goes to V- ? There is some writing/labeling but nothing stands out as a marker for one side vs the other.

Edit: OK I did some wiki-googling, and now know there should be some kind of mark indicating the cathode. Also learned that TVS diodes are installed reverse-biased, so that means I connect my cathode to V+ and my anode to V- correct?
Click to expand...

It depends if it's a uni- or bidirectional diode. Since the motors reverse I'd suspect it's a bidirectional diode though, so direction shouldn't matter. Do you have the part number so we can look up the data sheet? Normally there will be a silver ring on the - side if it's directional.

Edit: just found the part number above, looked up the data sheet - and the "CA" indicates that it's bidirectional. Polarity should not matter at all.
 
eKretz
It depends if it's a uni- or bidirectional diode. Since the motors reverse I'd suspect it's a bidirectional diode though, so direction shouldn't matter. Do you have the part number so we can look up the data sheet? Normally there will be a silver ring on the - side if it's directional.

Edit: just found the part number above, looked up the data sheet - and the "CA" indicates that it's bidirectional. Polarity should not matter at all.
Click to expand...

OK Thanks!

Once I get it installed I will check other components like you recommended. I have been looking up how to test transistors (paired diodes) both PNP and NPN.
 
Here is a pic of the board. I have been taking readings and will document and post. I have many questions on what I am reading on the transistors.


uc


uc
 
Last edited:
Here are my testing results (board not powered) of the SOT23 transistors.

https://drive.google.com/file/d/0B3fpxwOE8wVdU2ZXWVJQWDdFTzA/edit?usp=sharing

I don't see anything shorted, and I think all the opens are where you would expect. Let me know what you think.

I have traced down the white papers on the three different types of SOT23's that are there. A7, 04, and 2A. I looked them up on NXP/Phillips. I will update the spreadsheet of which one is which type if that will help. Also on there I have readings of other diodes and the large resistors on the board.

Are U1 - U4 transistors also? And what about M1 and M2 right next to the 24VDC power socket (J7)?

Update: Added the transistor type/markings to the spreadsheet
 
Last edited:
Can you post the pictures here? Apparently I haven't got permission to see whatever you have at that link.

U1 - U4 - can you post what the markings are, I can't quite make them out. The M components are your drive MOSFETs.
 
eKretz
Can you post the pictures here? Apparently I haven't got permission to see whatever you have at that link.

U1 - U4 - can you post what the markings are, I can't quite make them out. The M components are your drive MOSFETs.
Click to expand...
Sorry, you should be able to access now, I forgot to enable sharing on it. It is an excel file, but if need be I can make a pic of it and post.

U2 and U3 say V08 on them. I have learned that the letters are often country of origin tags (I left them off the other transistors listed) I assume it may be the same here with the V.

U4 says LM393L

I guess there is no U1, I thought the MOSFET next to/obscured by the resistor at the bottom was U1. It is actually M5.
 
Last edited:
It would probably be a bit easier to work with if you made a superimposition photo with the numbers you got when you measured at the end of leader lines going to the component legs you measured, but let me see what you've got.
 
eKretz
It would probably be a bit easier to work with if you made a superimposition photo with the numbers you got when you measured at the end of leader lines going to the component legs you measured, but let me see what you've got.
Click to expand...
I can do a pdf with comment boxes linked to each if that works. I hope you can follow my SOT23 pinout "key" and how I took the measurements.

I suspect the MOSFETS next to the TVS diode (M1 and M2), especially since on power up the motor is stuck pushing in one direction, and it looks like they are responsible for switching the voltage polarity to the motor. How do I test those? If it is too tricky/involved I may just buy two new ones and replace them, they can't cost that much.
 
Last edited:
Plug your board back in with all the connector and power up the wheel while checking the voltage at the pins (Neg lead to ground, Pos lead to each pin in turn). You'll probably need to do it several times to check all the pins since the wheel will be spinning. Check all the pins of both FETs while it is spinning for cal, then check again when it is trying to spin the opposite direction but can't due to the failure. We can then compare the two and see what's going on. Probably either the FET isn't switching properly or its not being switched on.
 
BTW did you say your wheel spins one way then shuts off or it keeps straining against the stop to turn in the first direction? I forgot if you said already.
 
eKretz
BTW did you say your wheel spins one way then shuts off or it keeps straining against the stop to turn in the first direction? I forgot if you said already.
Click to expand...
It keeps straining to go the same direction even though it is at the mechanical stop. I know the encoder wheel is ok, so my thought is the normal logic reads when the encoder wheel stops spinning as it's first calibration end point, then sends the signal to change directions. Since it does not switch, or even stop, that leads me to believe both mosfets may be bad. If only one switched from V+ to V- (or vice versa), then the 0V potential across the motor should make it stop. Of course I am making many assumptions there. I guess the other SOT23s that may be responsible for current switching that "tells" the mosfets to switch could be bad, or the logic chip itself got zapped.
 
This is what I have measured so far the foil layer regions indicated voltage as well. This is how the motor is stuck. Powering up and with calibration start, as far as I can tell nothing changes in this region voltage-wise.

Note: I took these board pics when the TVS diode was out, the new one is in place.

uc


I have done some research on the M1 and M2 MOSFETs. M1 is a 9977GM (dual n-channel mosfet) and M2 is a 4951GM (dual p-channel mosfet).
 
Last edited:
Also the sensor reading or sending the signals to the micro controller may not be sending its signal properly but that is probably less likely. The FETs looked like HEXFETs. Which are dual channel as you found out. There will always be p-channel and n-channel FETs when you have a motor drive as they work in conjunction.
 
Hmm, none of the other pins on the FETs have voltage? There should be voltage on at least one of the pins to switch the gate, as one FET is obviously on. Did you find the datasheet for the FETs? It would be helpful to have the pinout. What voltages are you getting on the transistor pins that are attached to the traces going to the FETs? (The ones you have the yellow lines following the traces to).

It's pretty unlikely that the microcontroller was damaged.
 
eKretz
Hmm, none of the other pins on the FETs have voltage? There should be voltage on at least one of the pins to switch the gate, as one FET is obviously on. Did you find the datasheet for the FETs? It would be helpful to have the pinout. What voltages are you getting on the transistor pins that are attached to the traces going to the FETs? (The ones you have the yellow lines following the traces to).

It's pretty unlikely that the microcontroller was damaged.
Click to expand...
In my post above with the pic and the markings, the text at the bottom with the model/markings of the FETs are links to their data sheets. They are matched pair gates (side by side) with the (redundant) drains/emitters facing in toward each other, two from one gate on one plane and the other two from the other gate on the other plain. If I am looking at this correctly, I would expect to see gate voltage (24v) at only one of the p-channel chip and also see it at the opposite n-channel, causing the current to flow in to one plane, through the motor and out the other plane like this:

uc


Instead, we have both p-channel gates and sources reading 24V, and 0V at all of the n-channel gates and sources. I would expect that (should the fet be intact) to send +VDC to both sides at the same time, causing a 0V potential, and no motor action. So I am wondering if the p-channel (M2) is bad and shorting and I am reading that voltage back through the short. At the same time though, that does not explain the lack of gate voltage at M1. I will test further up the line from both fets and update the posted pic above.

Edit: I updated my diagram above to correct the labelling of the Source terminals (was previously "C" for collector.)
 
Last edited:
Oops. Missed the links. The weird thing is that you have 24V at the drain but no 24V at the source nor any gate voltage. The MOSFET is on but by all indications it shouldn't be. What are you using for your ground? I.E. where are you putting your negative lead when measuring voltage?

Edit: never mind, I didn't see the 3rd 24V arrow going to G and S on M2. Thought that said 0V. I'll take another look shortly.

BTW did you just accidentally label source with a C because you have been looking at too many SOT23s? :D
 
eKretz
Oops. Missed the links. The weird thing is that you have 24V at the drain but no 24V at the source nor any gate voltage. The MOSFET is on but by all indications it shouldn't be. What are you using for your ground? I.E. where are you putting your negative lead when measuring voltage?

Edit: never mind, I didn't see the 3rd 24V arrow going to G and S on M2. Thought that said 0V. I'll take another look shortly.

BTW did you just accidentally label source with a C because you have been looking at too many SOT23s? :D
Click to expand...
Yeah, I have collectors and emitters stuck in my head from that.:lol:
C's non-withstanding, is my logic correct on how I think that should be operating in my sketch above?
Just to answer your question anyway, my black lead is on the negative/black pin of the back of the power jack.
 
Last edited:
Yes it sounds like you have it right. One quick thing you could check is with power off if you have any shorts between sources and drains.
 
eKretz
Yes it sounds like you have it right. One quick thing you could check is with power off if you have any shorts between sources and drains.
Click to expand...

I had just put this as an edit to the above, but then saw your post so I moved it here to make sense (you and I had the same idea at the same time):

I just did some testing, and I think I have found (the next) source of the problem. I think both M1 and M2 are bad. I am reading short in diode mode (and single digit ohms in resistance mode) across the FETs. It is the pair opposite to the ones I highlighted in green in my drawing, i.e. the one on M2 showing 24v at the drains and it's opposite on M1. That must be the pair that were active when the TVS went.

Could it be that the TVS diode they used is rated too high to clamp down before it can prevent damage to other components, or that the ones they used were poor quality/tolerance, or that even with a TVS diode in circuit there is always a risk of damaging other stuff?
 
Do you have the capability to remove the FETs and check them out of circuit? If not you'll just have to try swapping them I guess. They shouldn't cost much. When the TVS failed it may have let a little spike through - it doesn't take much to bugger an FET. You still need to check down the line and see if there are any other anomalies too though. This is sometimes difficult in-circuit unless you have a healthy wheel's PCB to compare voltages to. There are a lot of things down the line in parallel or series that can affect multimeter readings - so two identical BJTs may show very different readings in-circuit but read exactly the same out of circuit.

The TVS can only protect the circuit when it's working. It's supposed to clamp and release almost instantaneously once the voltage drops. once it's failed all protection can go out the window.
 
Last edited:
eKretz
Do you have the capability to remove the FETs and check them out of circuit? If not you'll just have to try swapping them I guess. They shouldn't cost much. When the TVS failed it may have let a little spike through - it doesn't take much to bugger an FET. You still need to check down the line and see if there are any other anomalies too though. This is sometimes difficult in-circuit unless you have a healthy wheel's PCB to compare voltages to. There are a lot of things down the line in parallel or series that can affect multimeter readings - so two identical BJTs may show very different readings in-circuit but read exactly the same out of circuit.

The TVS can only protect the circuit when it's working. It's supposed to clamp and release almost instantaneously once the voltage drops. once it's failed all protection can go out the window.
Click to expand...

Testing out of circuit how, with just the multimeter again or by setting up a test rig to test the gates and such?

Yeah, I plan on checking the circuit more (after the FETs are out). Even though they appeair to be bad, I don't like the looks of the all 24V hitting M2 and 0V on M1, there should be some opposed switching in my mind. I will get back to here once I have done more work.

Thanks for your help and guidance! :cheers:
 
Just double check that the short is definitely at the FET and not somewhere else in-circuit to make sure.
 
Hey Guy's,

I'm from the Netherlands, new on this forum and following your discussion about the circuit board, want you to know that I have problems likewise with my board.
You're latest respond is from a few week ago so just want to know if there is any success already by replacing the diode.
Shame that Logitech can't deliver a new board or even give an adress for repairs, after te warranty period of 3 years. It becomes simple waste after it goes to pieces.
Fortunately I have 2 other wheels, so I do have spares. Looking world wide for a new board without success so now I'm interrested in a repair.
 
Last edited:
The diode solved part of the problem, but it seems that there's still an issue with the switching. We will have to wait for a report on his findings.
 
Sorry for the delays, things have gotten kinda busy for me, so my progress is slow. So far I have traced out (most of) the circuits in that area of the board (and mapped them on a pic of the board). My next step is taking voltage readings at all points and then upload the combination of the two for you to look at. Just trying to figure out the best way to squeeze the voltages on the drawing as well.

I did remove the MOSFETS M1 and M2 and confirmed they're bad, read shorts just like on the board. That was very tricky, and I did end up tearing off one of the solder pads in the process. :banghead: :nervous: I need to work on my patience...:D

No worries though, as the trace leads to a nearby resistor that I can just solder a jumper wire to. I hope I find the final issue soon before I destroy it. :lol:

My soldering skills are not cut out for such small work under a magnifying glass. I did buy a new iron with built in LEDs and different tips to use (chisel tip as well) that would have been handy taking those FETs off!

Anyway, I hope to have the voltage reads done soon and posted here.

Oh yeah, I also noticed after taking the FETs off, the voltage feeding the FET gates and sources is still the same, 24V across both sets on M2 and 0V on both for M1. I am hoping we find a SOT23 that is stuck/bad and that is the final piece of the problem.

Will report back ASAP.

Thanks!

Edit: I updated my diagram in post #46 to correct the labelling of the Source terminals (was previously "C" for collector.)
 
Last edited:
Cor64
Hey Guy's,

I'm from the Netherlands, new on this forum and following your discussion about the circuit board, want you to know that I have problems likewise with my board.
You're latest respond is from a few week ago so just want to know if there is any success already by replacing the diode.
Shame that Logitech can't deliver a new board or even give an adress for repairs, after te warranty period of 3 years. It becomes simple waste after it goes to pieces.
Fortunately I have 2 other wheels, so I do have spares. Looking world wide for a new board without success so now I'm interrested in a repair.
Click to expand...

Like eKretz said, there are more issues. The diode was definitely bad, but I have also found that the two MOSFETS diagrammed above (M1 and M2) were also damaged and need replacing. I am now tracing out more of the area and reading voltages to find other problems as the voltages hitting the MOSFETs are incorrect, and likely another transistor or two is bad further up the line.

I was hoping at the beginning of this to be able to find that the diode fixed it all and post a guide for others to follow to fix their wheels as well. At this point I don't think that will happen as the territory I am venturing into is not for most people, and I myself would normally find it a better idea to spend the $125 (to get another one) instead of all this time. But I am taking this opportunity to use this as a learning experience as I have always had a basic knowledge, and great interest, in electronics. Since eKretz is so gracious with help and information, I could not resist the opportunity.

Having said that, if all goes well and I do get this fixed, I will do my best to summarize the process here should anyone be as adventurous (or desperate ;)) as I.

:cheers:
 
Last edited:
Will wait and see and great respect for your tenacity, both of you.:bowdown::bowdown:
Off the record, still wondering why Logitech isn't prepared to help us with a new board. By the way, I saw on your pictures that your board is a revison 001, same as mine. As I wrote before I do have more wheels and they have boards with revision 003. Does that mean a construction or design failiure in the past or just an improvement?

Logitech should order boards by their supplier only for people like us.
Wish I had an adress of their supplier in China.
Hope to hear from you soon.

Cheers, Bottoms up!
:cheers:
 
It could be an improvement or it could be as simple as some component changes due to availability or cost issues. Even something like a track routing change to put off less electrical noise, etc.
 

Latest Posts

Back