[MrMetric]

So, I have some interesting news on the 160V drive. I've moved to a test bench that has only a 12VDC supply and an encoder being hooked up to the drive. I pulled the heatsink off the drive and probed the differential receiver while turning the encoder. Interestingly, *neither* input is toggling (although I can see the toggling on the EXT pass-thru connector. So, either whatever is between the connector and the 26C32 receiver is bad or the receiver itself is sinking the input. This is the next area of investigation but it is absolutely the reason that the amp is not recognizing any turning of the encoder. And that is progress.

Of course, all of this begs the question of *how* the current state came about. There is no visible damage to the board or its components. No Char Broilers there, or even crispy looking stuff. Strange. The other good news is that the manufacturer is in contact with me now so I'm hoping we can resolve the (apparent) noise issues on the 80V drives. Fingers crossed.

[Larken]

Can you measure what the voltage is on each of the 4 encoder inputs with the encoder unplugged ? What voltage are they sitting at ?

If your using a singe ended encoder and you say it doesn't matter if the unused pin just sits unconnected, then what is the point of a differential encoder ?
Most encoder inputs have a pull up resistor to help the encoder drive the long cable, also there usually is a RC filter before the receiver chip, rarely would it just go straight into the chip.

If there is a pullup on a unused pin on a differential receiver, it won't work because its expecting the 2 wires to be a compliment of each other HL or LH, but if its unconnected you get HH and LH, so why would it work ? On the 75157 driver (like i use) you need to have the ununsed input sitting at 1/2 voltage (2.5 volts). (You should try this before probing the board, as you may accidently short something out )

It also could be a bad solder joint, but examining with a magnifier will rule this out.

Larry

[MrMetric]

I'm not in front of the setup right now but I think I can answer some of your questions...

First, I'm actually using a different encoder on the bench (BEI) than I am using on the machine (US Digital). I'll have to measure the floating voltage of each of the six encoder inputs (I'm counting the index as an input) but, yes, I agree on the comment regarding the filtering. The VSD-E's encoder input definitely is not connected directly to the differential receiver. It is connected directly to the pass through connector (used to pass the encoder to a Mesa board in my case.... but that is not connected right now). Although not asked, the EXT is showing as a normal TTL levels for low and high.

Regarding the need for a pull-up/down resistor for the floating leg of the differential input. I don't know what the VSD-E looks like yet so I can't really comment directly other than to say that the advised position within the Granite manual is that the differential side should be left floating (but, to avoid egg on my face, I will confirm that again just to be sure... BTW, I wouldn't mind egg on my face right now! ). However, the GD position does bear out in measured truth though. The two drives I have been fighting with have different problems. One can definitely "see" the encoder while the other cannot. So, for the purposes of determining what a "normal" behavior would be for the encoder, I confirmed that the signals do toggle at the differential receiver input, and output, with a floating leg. This confirmation was done on the drive that can see the encoder, of course, but which has very poor connection stability to the USB FTDI cable. When I try the same test with the drive that has a rock solid encoder count of 0, the signals do not toggle at all on the differential encoder input/output.

I'm pretty careful with the scope, but accidents do happen, I agree. I shall check the voltage when floating and let you know. It is a good suggestion, thank you... I can tell you (from memory) though that the inputs of the differential receiver are at TTL levels on the inputs. If I remember correctly, the unconnected leg is at logic 0.

As for the bad solder joint, I've used a pretty good loupe and can't see anything. Of course, that isn't a definitive test because something could still be hiding. A bad solder joint that is shorting should have been caught during the manufacturing process though. I'm hoping Tero will give me the input circuitry so that I can better determine a course of action and, if it comes to it, what to remove to get a better idea of how to proceed. Fortunately I do have good desoldering equipment, although I'm not really suited for SMT work. Where did all the DIP sockets go with their nice through holes!

I'm more familiar with the 157 chip, having used it a long time ago in another life, but the 26C32 is something foreign to me. Even there, though, I was always using the chip in true differential mode so the single ended mode was not something I really cared about. The project was a complete system, not something that was used by a myriad of different consumers such as the servo amps we are talking about.

[MrMetric]

Larry: I got a chance to measure the floating voltages and they are interesting....

80V drive (USB connection issue but definitely reading the encoder): The (+) side is 4.69V, the (-) is 2.19V
160V drive (unable to read encoder): Both the (+) and (-) are at 0V

[on edit]
I got a snippet of the schematic. From the input pin to which the encoder is connected, here is the schematic:

(+) - 2.2K pullup to +5V, then a series 2.2K resistor that connects to the differential amp. Also connected to the input is a 100pF capacitor to ground.

(-) - same as above except that the initial pullup resistor is not present. (not present on the snippet but found in the manual's equivalency circuit, are two 30K resistors, one pull-up and one pull-down, present on the (-) input to the differential receiver.

I'm a little perplexed where, on the non-counting 160V unit, the inputs are being pulled to ground when the input is not connected to the encoder. Even if the 100pF capacitors were defective (seems highly unlikely that they *all* are anyhow), or the 26C32 were defective and shorting the inputs to GND, then you'd have a voltage divider with 2.2K on both sides, resulting in a 2.5V float. I have to go back and confirm that when the encoder *is* connected, the encoder still toggles versus being pulled down.

Alan

[MrMetric]

So, I have news on the 160V supply, and it is very strange. The series 2.2K resistors all seem to have failed (and this includes on a second board that has a non-function, but irrelevant problem for this test, CPU). There are two 4 resistor networks that comprise the needed 8 series resistors (A, B, IDX, HOME... all differential). The resistors are all open circuit, not 2.2K.

The chances of this four components on two different boards failing due to manufacturing defects is statistically impossible. *Something* caused them to fail. Oddly, there is NO visible damage to the resistors, and no damage to the board indicative of heating. Very strange. The obvious thing would be that there was a voltage issue with some encoders, but then you'd expect components that were burned up... Strange.

[Montabelli]

I'm curious, did you acquire these drives new, or used?

[MrMetric]

LOL. If these were new, I'd have shipped them back for replacement a long time ago... Here is the history... Several years ago, I purchased four drives and motors, all used. One was described as known to be bad while the others were described as being fine. Two of these are 80V and one is 160V. I knew the seller indirectly and had no reason to believe there was anything amiss. Yes, caveat emptor. That said, I'm not made of gold and I took a gamble that was necessary to be able to complete the eventual project. Drives should be pretty reliable so I don't feel terrible about that. And if this were a poor decision, then we should all be looking at our purchases as a one way tunnel... Throw the money away because there will be no market for the item if we decide to sell it. I don't really believe in that.

I'm now (thankfully) in communication with Granite. They have said that they believe the drives are working (well, not the 160V one) based on the description. However, there have been some filtering caps added to later revisions that I need to add to my drives. I'll be doing that in the next week or so. I'm a little surprised why I'm having *so* many problems thought with the GDTool app. I can see *some* issues that would cause a rev but mine are almost constant (hence the frustration). But, be that as it may, I'll make the changes and see what happens.

As for the faulty 160V drive... I think we are on the right path here. I'll never be able to tune a drive that doesn't have the ability to read its inputs. If the 2.2K resistors have failed then those need to be fixed (I hate working on SMD stuff though... sigh). It seems odd that they acted as fuses without any visible damage but the probability that all of them just failed is stratospheric--something happened!

The previous owner disclosed that other 160V drive failed because he provided HV into the LV input. Clearly there was severe damage to the board (expected) there. I replaced the Freescale processor, and some other components in a Hail Mary that it might fix the problem (Granite advised me on what generally fails when this is done). Sadly, the drive is still non-op

[MrMetric]

I removed the suspected bad resistor networks from the 160V board and, sure enough, they were faulty. That would explain the inability to get the signal into the gate of the receiver. It is yet to be determined if the receiver is also bad and testing it is a little difficult right now. So, I'll replace the resistor packs and just try it again. Unfortunately, I don't keep any SMD stuff on hand so it will be the end of the week before I get those (and I just ordered some receivers too, just in case).

I setup the 80V VSD-E drives on a bench and utilized a much lower power servo (18VDC). When I operate with that, the amp works flawlessly. Therefore, it appears as though the 80V units are operational now, I just have some serious noise issues that are causing problems. Granite advised me to install two capacitors to help mitigate these issues, both of which are in transit now too. I may also have to replace the servo leads because those are not shielded. I tried using the original leads and perhaps that is just a bad idea.... I have to find a decent source for about 50 feet of 12GA shielded wire (more than I need, but I like to have spare stuff). Suggestions?

Anyhow, things look like they are moving forward and in the correct direction. This is a huge correction from how things were going before, so I'm happy. Next week I hope to have more updates (of a positive nature).

[MrMetric]

I thought I would update the thread on what I have learned/found....

First, I confirmed that the 160V did, in fact, have a hardware problem. The previously noted analysis was accurate in that the 2.2K series resistors and some of the 2.2K pullup resistors were faulty. They showed no signs of being burnt up, but after removal, they were confirmed as open circuit. Unfortunately, repairing these was a bit of a bear. I have thru-hole desoldering equipment but no SMT rework tools. There is limited space between the one side of the resistor packs and the connectors. This led to some frustrating moments, but eventually I was able to get the new packs soldering in and the boards were confirmed along with the two 80V units.

Confirmation of proper operation was performed through bench testing. The motor was powered by a lab grade bench supply providing 18V (the most I could easily get). Operation was flawless, without *any* communications problems with the USB interface, etc. So, that is *fantastic* news as it gives me confidence that the drives working (something I really didn't have before).

Per the manufacturer's recommendation, I have added two filter caps that were not present on my boards. One of these will be irrelevant because I am using the original wiring which is unshielded wiring. The added cap was between FG and ground, but my FG isn't connected to the shield. However, i added the cap in case I have to rewire my motor leads with shielded cables. The second cap is between two of the serial leads, so I'm guessing/hoping that it might clear up some of my issues. I also will likely order some EMI cores to help mitigate any spurious noise issues.

So... very good progress. I'm feeling *much* better about my starting point now as I embark on my 2.0 attempts after the complete reboot, so to speak. I'll keep everyone posted if they are interested.

[MrMetric]

Here is the latest update...

As previously mentioned, I repaired the 160V drive and added the filter caps to all of the drives. I also added EMI filters on both ends of the servo motor leads. On the motor side, I could make a full turn around the core; that was a Laird LFB250150. On the amp side I used one of those plastic clip on leads, so no wrap; that was a Steward HFA150066-OA2. After doing all this, *all* of my connection problems with the FTDI USB interconnect went away. It is rock solid... what a relief! I can't say that I'm completely happy with my tuning results. My graphs don't quite turn out the way I was expecting (i.e. are not tracking as closely as the documentation on the Granite site). In fairness, I'm still getting my sea legs on understanding how the different parameters affect the system. When I get the graphs to superimpose better, then the drive is "louder" (almost like sand), which just feels wrong. So, I'm not sure where the happy compromise is.

However, these tuning issues are completely separate from the inability to connect to the driver, which was the topic of this thread. Therefore, I will likely open a different thread for the tuning questions sometime in the future. Right now I think I'll concentrate on getting the rest of the machine setup, the I/O mapped in the controller, etc. I can always tune later as long as the amplifiers are tuned well enough that they are stable (which they are).

One thing I can say is that the setup is *so* much quieter than the old amps that were in there. The chopping frequency is high enough that I don't hear it anymore. Getting rid of that squealing was one of the main reasons I opted to move to new amps, so I couldn't be happier about that!

So... progress.... That is a great thing. Tero/Granite came through with some pivotal information to help me diagnose the hardware issues on the encoder. He also game great information on the filter components that needed to be added. So, I'm thankful for that and happy everything is moving in the right direction again. Also, a BIG thank you to everyone here that helped with ideas and suggestions. It is very much appreciated.

Alan