Problems, problems

I installed the PCGU1000 and its software (all the most recent release) on an XP-SP3 machine. It’s giving me major problems. The software may connect to the instrument for one or two commands, but quickly it loses contact with the device. Often the device will continue to output the most recent waveform. Cycling the USB connection helps in maybe 1 out of 10 tries. Cycling the power to the device is usually necessary to get it to communicate again, but even then, it loses the connection within two or three new commands. Needless to say, this is pretty frustrating. Does anyone have any ideas? When the device works, it works really well.

Hi,

Have you looked at this post?
http://forum.velleman.eu/viewtopic.php?f=11&t=2514

I hope this helps.

Dang. No. I hadn’t seen that. Thank you. Do devices that have been modded that way still tolerate the full range of USB power supply voltages? The low voltage permitted by the USB 2.0 standard is 4.4V. Adding that LED in there seems like it would adversely affect its ability to run at a supply that low. Maybe I’ll try something like a 1n4148 instead of an LED and see how that works.

It’s kind of interesting that this is a device that requires external power, yet still draws power from the USB – no 9V power supply connection is needed for the device to communicate and identify itself to the computer.

I think if you make the modification that the Velleman team has pointed out you will be very happy with the unit.

I’m not sure but changing the opticoupler may cause other problems.

The only thing I can tell you is this fix works very well.

Once again I hope this helps.

[quote=“GeneralEclectic”]Dang. No. I hadn’t seen that. Thank you. Do devices that have been modded that way still tolerate the full range of USB power supply voltages? The low voltage permitted by the USB 2.0 standard is 4.4V. Adding that LED in there seems like it would adversely affect its ability to run at a supply that low. Maybe I’ll try something like a 1n4148 instead of an LED and see how that works.

It’s kind of interesting that this is a device that requires external power, yet still draws power from the USB – no 9V power supply connection is needed for the device to communicate and identify itself to the computer.[/quote]

No need to worry, the opto-coupler’s recommended input current in 5 to 15mA–even with a 4.4V USB line (I’ve never seen one that low, most are a bit high) the flow would be (4.4 - 2.0) / 330 = 7.3mA–well within the coupler’s recommended range.

It draws the small amount of power from the USB port to operate in opto-couplers and provide complete electrical isolation of the PC and the generator output–this is a good thing…

No, actually, it’s a lot worse than that. It’s running much closer to device margins than you think. You neglected the Vf of the LED in the optoisolator. If you include that, it’s 4.4-2-2V = .4V/330ohm. That’s on the order of 1ma. That’s WAY too low IMO, and makes no allowance for unit to unit Vf variation and temperature effects, putting you into an ugly spot with respect to the current transfer function of the isolator. If I decide to mod this, it’s gonna be a 1N4148 or 1N2002, not an LED. One of those diodes will get you about .5V-.6V at these current levels. An even better choice might be just another resistor of maybe 220 ohms or so. It would cut the peak current without introducing so much additional voltage sensitivity as a diode with its fixed forward voltage does.

One can debate the design philosophy of having a device that “reports in” even though it’s actually unpowered. You can make a pretty good case for putting the USB processing on the internally-powered side of the optoisolators. IMO, that would lead to a little less potential confusion as to the state of the device itself, vis-a-vis the power supply. (That’s the same power supply that I’ve found to be extremely flaky because of that dreadful universal plug design.) It would have been much better had the designer used a dual-color LED for the status light. Red for “power” and Green for “power and communication” would have been far more user-friendly.

After playing with this thing for a little while, I’m just amazed at its frequency accuracy. The voltage accuracy looks excellent too, much better than anyone has a right to expect in a $200 device. It’s a shame that it’s hampered by a somewhat clumsy and apparently undocumented UI. I’m still trying to find a real user manual that explains how to use all the capabilities it offers.

General,

The only thing I can tell you is that the mod works very well.
Since I have done it to mine it has never failed to connect.

Have you looked in the help section of the GUI for the PCGU1000?

I have also noticed that the admins and users of this board (like CliffyK) are a large help.
I have learned a lot from all of them just from lurking around and reading others post.

Most of the time I get answers before I ask.

I hope you have a happy holiday.

You are correct, I discounted the internal LED, nonetheless and as reported by others the mod works great–of the 6 or 7 PCs and notebooks available to me it connects solidly every time.

As to the documentation, I agree it is a bit sketchy (and to my eye a bit too “cartoony” for a serious piece of test equipment) however I find the instrument’s operation to be quite straight-forward–I’ve been using it for quite some time now and would be pleased to address any specific issues.

-cliff-

PS: I haven’t forgotten about packaging up the latest source code for my front-end, we’re not at home this week…

[quote=“GeneralEclectic”]No, actually, it’s a lot worse than that. It’s running much closer to device margins than you think. You neglected the Vf of the LED in the optoisolator. If you include that, it’s 4.4-2-2V = .4V/330ohm. That’s on the order of 1ma. That’s WAY too low IMO, and makes no allowance for unit to unit Vf variation and temperature effects, putting you into an ugly spot with respect to the current transfer function of the isolator. If I decide to mod this, it’s gonna be a 1N4148 or 1N2002, not an LED. One of those diodes will get you about .5V-.6V at these current levels. An even better choice might be just another resistor of maybe 220 ohms or so. It would cut the peak current without introducing so much additional voltage sensitivity as a diode with its fixed forward voltage does.

One can debate the design philosophy of having a device that “reports in” even though it’s actually unpowered. You can make a pretty good case for putting the USB processing on the internally-powered side of the optoisolators. IMO, that would lead to a little less potential confusion as to the state of the device itself, vis-a-vis the power supply. (That’s the same power supply that I’ve found to be extremely flaky because of that dreadful universal plug design.) It would have been much better had the designer used a dual-color LED for the status light. Red for “power” and Green for “power and communication” would have been far more user-friendly.

After playing with this thing for a little while, I’m just amazed at its frequency accuracy. The voltage accuracy looks excellent too, much better than anyone has a right to expect in a $200 device. It’s a shame that it’s hampered by a somewhat clumsy and apparently undocumented UI. I’m still trying to find a real user manual that explains how to use all the capabilities it offers.[/quote]

Hi Guys. Thanks for the feedback, as always.

For the moment, I’m not going to make any changes to the input optoisolator circuit. The device is now plugged onto a PCI-USB accessory board that seems to make it happy. I think that part of the problems I blamed on the USB were actually due to that miserable power supply. THAT thing is crap, and I’m sort of ticked at the Velleman people for putting something so problematic with this otherwise surprisingly good instrument.

The problem with the power supply has to do with the two rectangular pins making intermittent contact with the localizing plug adapter internally. I’m guessing that there’s insufficient spring force on the internal contacts, or perhaps one or both of them was damaged prior to adapter final assembly. Regardless, that thing is constantly flickering off and on. It’s a good thing that it has a pilot light on it. It still may, one day, go sailing across the room. Maybe the best thing to do for now is just substitute another 9VDC wallwart with an appropriate DC plug grafted onto it.

[quote=“GeneralEclectic”][color=#8080FF]Hi Guys. Thanks for the feedback, as always.

For the moment, I’m not going to make any changes to the input optoisolator circuit. The device is now plugged onto a PCI-USB accessory board that seems to make it happy. I think that part of the problems I blamed on the USB were actually due to that miserable power supply. THAT thing is crap, and I’m sort of ticked at the Velleman people for putting something so problematic with this otherwise surprisingly good instrument.

The problem with the power supply has to do with the two rectangular pins making intermittent contact with the localizing plug adapter internally. I’m guessing that there’s insufficient spring force on the internal contacts, or perhaps one or both of them was damaged prior to adapter final assembly. Regardless, that thing is constantly flickering off and on. It’s a good thing that it has a pilot light on it. [/color]It still may, one day, go sailing across the room. [color=#8080FF]Maybe the best thing to do for now is just substitute another 9VDC wallwart with an appropriate DC plug grafted onto it.[/color][/quote]

That’s just what mine did (fly across the room) the same day I got the unit, I soldered a surplus line cord to the pins–that and a bit of heat-shrink tubing, and I had a solid connection.

I have never seen one of those “international” wall-warts with the clip-in adapters that was worth a crap…

Funny!

I was considering the soldered-line-cord trick, after seeing if modified fastons might be made to work on those thick prongs. I was wondering if that small recessed metal prong in the corner is a tap changer of some sort on the input transformer, because the 120VAC adapters have a post that “engages” it, but the 240 adapters do not. Since you’re applying 120V to the power supply without any interaction with that post and it’s working, I guess it has a pretty wide input voltage range.

[quote=“GeneralEclectic”]Funny!

I was considering the soldered-line-cord trick, after seeing if modified fastons might be made to work on those thick prongs. I was wondering if that small recessed metal prong in the corner is a tap changer of some sort on the input transformer, because the 120VAC adapters have a post that “engages” it, but the 240 adapters do not. Since you’re applying 120V to the power supply without any interaction with that post and it’s working, I guess it has a pretty wide input voltage range.[/quote]

Toothpick, a rather robust one, jammed in the hole…