I cannot get any sensible impedance readings from an IMP measurement system. A 6R resistor reads around 30R, and a dead short around 5R. I have tried (and declared in the settings) both a 10R and 20R reference resistor with much the same result. I have also reloaded the 2.05 software to eliminate any rogue settings.
Is there anyone else out there still using this system or who has a good enough memory to recall whether there are any gotchas I might have missed?
Frequency response readings all look OK.
Is there anyone else out there still using this system or who has a good enough memory to recall whether there are any gotchas I might have missed?
Frequency response readings all look OK.
Hi Ardee,
I used Imp for several years, but now I use Praxis, so my memory on the Imp system is a very vague. However, for the problem you describe, I can think of two things that may be causing your problem;
1) Have you run the calibration procedure recently. To my knowledge, impedance measurements with all these computer based systems employ both left and right channel inputs of a sound card that are effectively connected to either side of the reference resistor. If there are any dicrepancies in the sensitivity between these two inputs, the resulting measurements will be in error.
2) For my setup, I recall that I had to use a small power amp to buffer the output when making impedance measurements in order to provide enough current to drive the low impedance loads.
I assume you are doing the same, so I suggest you check out your power amp to see if that is functioning correctly.
Hope this helps.
Peter
I used Imp for several years, but now I use Praxis, so my memory on the Imp system is a very vague. However, for the problem you describe, I can think of two things that may be causing your problem;
1) Have you run the calibration procedure recently. To my knowledge, impedance measurements with all these computer based systems employ both left and right channel inputs of a sound card that are effectively connected to either side of the reference resistor. If there are any dicrepancies in the sensitivity between these two inputs, the resulting measurements will be in error.
2) For my setup, I recall that I had to use a small power amp to buffer the output when making impedance measurements in order to provide enough current to drive the low impedance loads.
I assume you are doing the same, so I suggest you check out your power amp to see if that is functioning correctly.
Hope this helps.
Peter
Thanks, but the IMP uses both a calibration input which is connected to the output from the amplifier and a measurement input connected to the junction of the reference resistor and item under test. Like most systems used to measure impedance it just subtracts one input from the other. In my case, something is upsetting the arithmetic. The cold side of the amplifier output is grounded to the IMP, by the way, as the amplifier has an input transformer.
Hi Ardee,
I use the IMP frequently. The only thing I ever had a problem with was the input RCA jacks breaking the solder joints on the IMP PC board. This gave erroneous reading. A quick re-solder fixed the problem. You might want to check.
Just noticed that you said the amp has in input transformer. Have you tried another amp? Or, try using the IMP output directly (no amp in the circuit) with a 1k ohm ref resistor and a 1k test resistor.
I use the IMP frequently. The only thing I ever had a problem with was the input RCA jacks breaking the solder joints on the IMP PC board. This gave erroneous reading. A quick re-solder fixed the problem. You might want to check.
Just noticed that you said the amp has in input transformer. Have you tried another amp? Or, try using the IMP output directly (no amp in the circuit) with a 1k ohm ref resistor and a 1k test resistor.
Hello John K, and thanks for the suggestions.
Input jacks all seem soundly connected, and I did check ground continuity between them.
Using 1k resistors gives a reading of about 3k. Phase looks about right at 0° rising a bit at the HF end. What might be significant is that if I short out the reference resistor and do an electrical test with get-new-reference set, it shows a 0dB flat line. With the reference resistor in circuit it shows a flat line about 2.5dB down. I would have expected this to be either 3 or 6dB, depending on whether it is showing power or voltage. Another odd thing I notice is that, when doing Acquire-Repeat to set the input levels, if I keep turning the level up past overload the displayed level starts dropping. Could a missing bit cause this? I have never really liked the look of that ribbon cable on the Centronics connector.
Input jacks all seem soundly connected, and I did check ground continuity between them.
Using 1k resistors gives a reading of about 3k. Phase looks about right at 0° rising a bit at the HF end. What might be significant is that if I short out the reference resistor and do an electrical test with get-new-reference set, it shows a 0dB flat line. With the reference resistor in circuit it shows a flat line about 2.5dB down. I would have expected this to be either 3 or 6dB, depending on whether it is showing power or voltage. Another odd thing I notice is that, when doing Acquire-Repeat to set the input levels, if I keep turning the level up past overload the displayed level starts dropping. Could a missing bit cause this? I have never really liked the look of that ribbon cable on the Centronics connector.
Both probes are 47k, which is declared. I think that any problem with the probes would cause the probe 1 and 2 input levels to show differently when the reference resistor is shorted, but they don't.
I've tried different PC's, so unless there is a setting I have overlooked, I can only think this must be a hardware problem or that a "fast" (Pentium 2) PC is just too much for it. Using SlowDOS makes no difference, but this might not affect the printer port speed.
I've tried different PC's, so unless there is a setting I have overlooked, I can only think this must be a hardware problem or that a "fast" (Pentium 2) PC is just too much for it. Using SlowDOS makes no difference, but this might not affect the printer port speed.
Try this.
Connect the pulse output of the IMP directly back to the Cal input, (Probe 1). Do not use any resistive probes, just a cable.
Set the probe gain pot to the first mark (almost off).
Select auto_Measure.
Click just_Cal.
You should see a flat line with slight down turn above 10k Hz. The level should be close to 0 dB. If the level is too low, or too high adjust the gain pot appropriately.
On the IMP unit change the cable from Probe1 input to Probe 2 input. DO not change the pot gain setting.
Then go to the auto_Measure menu and click Elec.
You should see a flat response at 0 dB.
If you get this the IMP is working correctly.
One other question. What operating system are you running? I run the IMP on an older W95 machine. As I recall, it won't run on XP. Don't know exactly were the cut off was.
Connect the pulse output of the IMP directly back to the Cal input, (Probe 1). Do not use any resistive probes, just a cable.
Set the probe gain pot to the first mark (almost off).
Select auto_Measure.
Click just_Cal.
You should see a flat line with slight down turn above 10k Hz. The level should be close to 0 dB. If the level is too low, or too high adjust the gain pot appropriately.
On the IMP unit change the cable from Probe1 input to Probe 2 input. DO not change the pot gain setting.
Then go to the auto_Measure menu and click Elec.
You should see a flat response at 0 dB.
If you get this the IMP is working correctly.
One other question. What operating system are you running? I run the IMP on an older W95 machine. As I recall, it won't run on XP. Don't know exactly were the cut off was.
Hi Ardee,
I had a problem with my IMP in that the probe inputs would not work but the microphone input would. The input multiplexer (a 74HC4053) had failed, I think due to me inadvertently leaving the 47K resistors out of the probes. A standard 4053 was to hand and works OK. That is the only failure in 10 years of use. I use an old laptop running Windows ME and the software runs OK in a DOS window
Its great in combination with the DCX2496 you can quickly 'dial in' a crossover.
My blog:
http://consort3.wordpress.com/
I had a problem with my IMP in that the probe inputs would not work but the microphone input would. The input multiplexer (a 74HC4053) had failed, I think due to me inadvertently leaving the 47K resistors out of the probes. A standard 4053 was to hand and works OK. That is the only failure in 10 years of use. I use an old laptop running Windows ME and the software runs OK in a DOS window
Its great in combination with the DCX2496 you can quickly 'dial in' a crossover.
My blog:
http://consort3.wordpress.com/
Try a different power amplifier. IMP depends on the "cold" [-] side of the amplifier output being always at the same potential as the input ground/shield connection. Bridged amplifiers, or most amplifiers with transformer in their outputs, won't work. The ground from the "pulse out" has to be the same circuit node as what connects to the negative terminal of the speaker (or resistor). That's one of the changes in PRAXIS, the inputs are differential so that input ground doesn't have to be output ground, too. Also, it can work from a current measurement rather than from subtracting voltage measurements (which isn't so tolerant of errors).
Ditto on the 4053 maybe being bad. It will get damaged if it ever gets connected to a power amplifier without the resistors being in the probes. They don't cost much and are easy to replace, so its worth a try.
Ditto on the 4053 maybe being bad. It will get damaged if it ever gets connected to a power amplifier without the resistors being in the probes. They don't cost much and are easy to replace, so its worth a try.
Well, it looks as if my original problem was a portent. Everything was left on the bench untouched but, after powering up today, any attempt to measure gives the "IMP does not respond" message.
The IMP is within reach of the 'scope, but as I don't know what hardware response the software is looking for, diagnosis is difficult.
It is at this point that I say thanks to Bill Waslo for responding, and ask if you could suggest what hardware condition is likely to be missing.
I am not using any amplifier at this stage, by the way, just the 1k resistors suggested by john k.
Thanks.
The IMP is within reach of the 'scope, but as I don't know what hardware response the software is looking for, diagnosis is difficult.
It is at this point that I say thanks to Bill Waslo for responding, and ask if you could suggest what hardware condition is likely to be missing.
I am not using any amplifier at this stage, by the way, just the 1k resistors suggested by john k.
Thanks.
Ardee,
Don't give up.
I have had the IMP is not responding message pop up numerous time. I would suggest doing 2 things. First, disconnect the IMP power supply and ground the center pin to the outer ring of the socket on the IMP. Second, disconnect the IMP from the printer port and restart the PC. Then reconnect everything ans see what happens.
I don't know exactly what the cause is here, but I have suspected that the printer port interface gets corrupted and communication is lost. Just a guess, but making sure the IMP is powered down and the printer port cleared seems to correct the problem if it is not a hardware failure.
Don't give up.
I have had the IMP is not responding message pop up numerous time. I would suggest doing 2 things. First, disconnect the IMP power supply and ground the center pin to the outer ring of the socket on the IMP. Second, disconnect the IMP from the printer port and restart the PC. Then reconnect everything ans see what happens. I don't know exactly what the cause is here, but I have suspected that the printer port interface gets corrupted and communication is lost. Just a guess, but making sure the IMP is powered down and the printer port cleared seems to correct the problem if it is not a hardware failure.
Thanks John, you are right. If I power up the IMP and start the software before connecting the printer port, it works. I have had the IMP for years and I do now remember this happening once before.
Having got back to square one, I tried your suggested test. I get a straight line from probe 1 and just a high level mess from probe 2, so next step is to change the 4053. If that does prove to be the cause I'm fitting the 47k resistors internally!
To reply to other questions, I normally use the IMP with W98SE DOS, or FreeDOS. I would be reluctant to replace the system, as it does everything I want. The only major flaw it has is in not displaying an optimised full-range measurement by automatically combining both high and low measurements.
Having got back to square one, I tried your suggested test. I get a straight line from probe 1 and just a high level mess from probe 2, so next step is to change the 4053. If that does prove to be the cause I'm fitting the 47k resistors internally!
To reply to other questions, I normally use the IMP with W98SE DOS, or FreeDOS. I would be reluctant to replace the system, as it does everything I want. The only major flaw it has is in not displaying an optimised full-range measurement by automatically combining both high and low measurements.
What I do is to save the data files (FRD files) for the high and low samplining rates and edit them by hand to creat a merged file. I would do it in the IMP but the combined result has such poor resolution (15 Hz ) at low frequency it is not realy useful.
Sound like Probe 2 imput is bad, but again., hit the RCA jack PCB pins for Probe 2 with a soldering iron before changing a chip.
Sound like Probe 2 imput is bad, but again., hit the RCA jack PCB pins for Probe 2 with a soldering iron before changing a chip.
The 4053 is innocent, I'm afraid.
Once I had changed the 4053, I found that the reason for the messy elec reading at port 2 was that I had forgotten to set the cal_source to "none" Having done that both ports give a perfect result.
I am really at a loss to explain the odd impedance readings. Using a 1k reference resistor it still shows a 1k test resistor as just over 3k. With this 1k + 1k setup the difference between the port 1 and 2 plots is slightly under 3dB. I would be very grateful if someone could confirm that their IMP produces the same result. If it does, at least I will know the hardware is OK.
Once I had changed the 4053, I found that the reason for the messy elec reading at port 2 was that I had forgotten to set the cal_source to "none" Having done that both ports give a perfect result.
I am really at a loss to explain the odd impedance readings. Using a 1k reference resistor it still shows a 1k test resistor as just over 3k. With this 1k + 1k setup the difference between the port 1 and 2 plots is slightly under 3dB. I would be very grateful if someone could confirm that their IMP produces the same result. If it does, at least I will know the hardware is OK.
john k... said:
Have you done this, exactly? This assures that the only thing in the change in the IMP. If the signal level is not 0 dB there is some kind of internal problem.
I suspect that if this comes out -3dB you could adjust the gain so it comes out 0dB and then the impedance would also work. But I would be concerned about the problem.
I know you have probably done this but just making sure. Hard to help debug over the net.
John,
If I follow your procedure without any resistors, the ports give an exact 0dB match. The result is also exactly the same with the 47k input resistors in circuit and the probe level set higher.
To give a final all-clear to the hardware, though, it would be interesting to know what the difference in levels should be with the 1k + 1k resistor arrangement you previously suggested. If the imp is displaying voltage, I would expect the difference between the port 1 and port 2 plots to be about 6dB, as the two equal resistors should give half the voltage at port 2 that there is at port 1.
Does that make any sense?
If I follow your procedure without any resistors, the ports give an exact 0dB match. The result is also exactly the same with the 47k input resistors in circuit and the probe level set higher.
To give a final all-clear to the hardware, though, it would be interesting to know what the difference in levels should be with the 1k + 1k resistor arrangement you previously suggested. If the imp is displaying voltage, I would expect the difference between the port 1 and port 2 plots to be about 6dB, as the two equal resistors should give half the voltage at port 2 that there is at port 1.
Does that make any sense?
I ran some tests with my IMP.
When I do the loop back test with or without 47k ohm probes I get 0dB between channels. But when I do the 1k, 1k voltage divider I have to use the probes to get the correct result. My error if I gave the impression to do the voltage divider w/o probes. It should give -6dB for the level measured between the resistors.
Anyway, what you are telling me leads me to think the IMP is ok and the problem is elsewhere. The only other thing I can think of is the probeZ is not set correctly in the IMP softare (under steup).
Other than that I'm at a loss.
When I do the loop back test with or without 47k ohm probes I get 0dB between channels. But when I do the 1k, 1k voltage divider I have to use the probes to get the correct result. My error if I gave the impression to do the voltage divider w/o probes. It should give -6dB for the level measured between the resistors.
Anyway, what you are telling me leads me to think the IMP is ok and the problem is elsewhere. The only other thing I can think of is the probeZ is not set correctly in the IMP softare (under steup).
Other than that I'm at a loss.
Thanks for doing that, and for all your previous help. My IMP passes the 0dB loop back test but shows less than 3dB difference between probes with the 1k resistors in place. If yours shows 6dB, which is what I would expect, this should be telling me something, but I have no idea what that something is.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.