Re: Re: HP-339A problems
Don't change ANYTHING automatically! There lies divergence and ruin.
If the supplies are OK and ripple is low, leave them alone. If there are problems, look for a faulty or degraded component and change that. You have a 'scope, which is a good start.
I have fixed a number of these, currently got a 333A on the bench. Switches need attention.
bsgd said:
Don't change ANYTHING automatically! There lies divergence and ruin.
If the supplies are OK and ripple is low, leave them alone. If there are problems, look for a faulty or degraded component and change that. You have a 'scope, which is a good start.
I have fixed a number of these, currently got a 333A on the bench. Switches need attention.
1audio said:
Hi Demian,
What I mean is, I did the calibration procedure just like stated on the service manual, setting the full scale and 1/3 scale pots with the 339A internal oscillator set to 1kHz 3V. All measurements were made with a pair of Fluke 87-III DMM's, and they both agree on the readings.
After the calibration I could successfuly measure voltages in the 1-4 V, but very low voltages like 10mV have a 10% error. Also, if I set the oscillator to high frequencies, like 100kHz, voltage measurements get even worse: My Fluke says 3mV, my 339A says 8mV. At high frequencies, I also have problems with the oscillaor ouput: turning the level vernier (with the control set to 10mV) doesnt seem to alter the readings on my Flukes, but the HP meter does move indicating a change in the volatge of the internal oscillator.
I do have a scope if needed.
I meant that it should not be treated like a boutique audio setup with caps and resistors changed wholesale with majik components to "improve the sound."
The ripple specs are in the docs. If it meets those then look elsewhere.
You said "I do have a 'scope if needed". That frightens me since it should be the first instrument to use before any digital meter.
Your Fluke is a good one - is it one of the true RMS models? - but I would not believe ANY digital meter unless I was sure it was measuring only DC. Stick ripple or oscillations on it and it is a lottery.
I have a cheap Tek 5000 series 'scope which I know is good to 5%, and 2% with care, and I use it for all preliminary measurements. If it is in the park, then out comes the Keithley (in my case) for a final tweak.
cliffforrest said:
I meant that it should not be treated like a boutique audio setup with caps and resistors changed wholesale with majik components to "improve the sound."
The ripple specs are in the docs. If it meets those then look elsewhere.
You said "I do have a 'scope if needed". That frightens me since it should be the first instrument to use before any digital meter.
Your Fluke is a good one - is it one of the true RMS models? - but I would not believe ANY digital meter unless I was sure it was measuring only DC. Stick ripple or oscillations on it and it is a lottery.
I have a cheap Tek 5000 series 'scope which I know is good to 5%, and 2% with care, and I use it for all preliminary measurements. If it is in the park, then out comes the Keithley (in my case) for a final tweak.
Well, I do have a few boutique caps if needed hahahaha.. maybe some Black Gates? OK, I get your point, no careless replacing of parts.
Regarding the scope, I do have it, BUT, its a PC based scope. Its a Velleman PCSU1000. It works pretty well for what I need, but I never needed a scope for measurement in such precise instruments....
Yes, both my Flukes are True RMS. I like them pretty much as my previous DMMs were cheap and not high quality at all.
OK, in that case I would just finish with:
If your DVM gives a reading which doesn't seem to make sense, check the waveform before believing it and going on a wild goose chase.
A few 1% resistors (10R, 100R, 1K, 10K etc) make it easy to generate accurate low voltage DC references to verify.
If the 15V rail measures 15V on both meters (having 2 is a great help!) then a pot of 10K/10R will generate an accurate 15mV test voltage. You have to allow for DVM input Z, yada yada yada etc
If your DVM gives a reading which doesn't seem to make sense, check the waveform before believing it and going on a wild goose chase.
A few 1% resistors (10R, 100R, 1K, 10K etc) make it easy to generate accurate low voltage DC references to verify.
If the 15V rail measures 15V on both meters (having 2 is a great help!) then a pot of 10K/10R will generate an accurate 15mV test voltage. You have to allow for DVM input Z, yada yada yada etc
Hi Demian,
Yes, I pretty much agree with you on those things.
I have a number of 33x and 65x units I want to restore and possibly improve slightly. Any thoughts of a drastic improvement with these is a pipe dream. Those engineers knew exactly what they were doing.
Hi bsgd,
I think you made a major mistake when you performed a calibration on your 339A. I normally find that these are well in tolerance as long as no one else has been in there. The first rule is "HP / Agilent equipment generally holds it's calibration well. Twisting controls will be the very last step and only if you can prove it's out of tolerance.
To properly clean the switches in any test equipment involves care and attention. You must carefully apply the cleaning solution to a wood shaft "Q-Tip" (or similar) and very carefully rub the center contact rings. They will become shiny as you clean them. You will probably have to remove some of the cotton first. Take extreme care you don't get the cotton caught in a contact "finger" in case you bent it (Baaaad news!). You will get the hang of this. Once clean, rotate the control knob to get the cleaner on the inside of the finger contacts. Follow this up with a zero residue cleaner, plastic safe. Hold the instrument up-side-down so that nothing drips onto the interior. Do not get anything in a trimmer capacitor - ever!!!
After you clean the controls, you can make some meaningful tests.
Do check the power supplies for correct voltage and noise (if you can). Replace any parts with ones of the same values and physical size. That often means you will be using a higher voltage capacitor. These normally have better dielectric properties. Increasing capacitor sizes by any great amount may degrade the performance of these wide band instruments.
The 8903A has a higher residual noise than the 339A. It's more on par with the 333A/334A. These have auto-nulling. The 331A and 332A do not have auto nulling. The 332A and 334A have AM demodulator inputs, otherwise they are the same as the 331A and 333A. For normal audio work, the 333A or 334A would be good choices. The auto-nulling will track slowly drifting oscillators or filters. I have occasionally been able to get a deeper null in manual mode (= to 331A and 332A in this mode).
Your Fluke 87 meters are very good and do tend to hold their calibration ... except you can not depend on the higher frequency AC readings at all. An HP 339A is more accurate on AC than a Fluke 87. Why you ask? Simple. The person who calibrates a Fluke 87 knows that once you change the case, or reinstall it, the AC readings take a wander. The most accurate way to align an 87 is to use another top cover with access points drilled out. Do the adjustments and reinstall the proper case. Then check to see how far out they are. I kid you not. If a replacement cover was not used, consider the higher frequency readings as if they are vague suggestions unless you have before and after readings in you hands that prove the readings are correct. I can say with certainty that one that is right on is a rare piece. Later Fluke meters that use what is called "closed case calibration" would be expected to be in tolerance.
Have I caught everything you were wondering about?
-Chris
Yes, I pretty much agree with you on those things.
I have a number of 33x and 65x units I want to restore and possibly improve slightly. Any thoughts of a drastic improvement with these is a pipe dream. Those engineers knew exactly what they were doing.
Hi bsgd,
I think you made a major mistake when you performed a calibration on your 339A. I normally find that these are well in tolerance as long as no one else has been in there. The first rule is "HP / Agilent equipment generally holds it's calibration well. Twisting controls will be the very last step and only if you can prove it's out of tolerance.
To properly clean the switches in any test equipment involves care and attention. You must carefully apply the cleaning solution to a wood shaft "Q-Tip" (or similar) and very carefully rub the center contact rings. They will become shiny as you clean them. You will probably have to remove some of the cotton first. Take extreme care you don't get the cotton caught in a contact "finger" in case you bent it (Baaaad news!). You will get the hang of this. Once clean, rotate the control knob to get the cleaner on the inside of the finger contacts. Follow this up with a zero residue cleaner, plastic safe. Hold the instrument up-side-down so that nothing drips onto the interior. Do not get anything in a trimmer capacitor - ever!!!
After you clean the controls, you can make some meaningful tests.
Do check the power supplies for correct voltage and noise (if you can). Replace any parts with ones of the same values and physical size. That often means you will be using a higher voltage capacitor. These normally have better dielectric properties. Increasing capacitor sizes by any great amount may degrade the performance of these wide band instruments.
The 8903A has a higher residual noise than the 339A. It's more on par with the 333A/334A. These have auto-nulling. The 331A and 332A do not have auto nulling. The 332A and 334A have AM demodulator inputs, otherwise they are the same as the 331A and 333A. For normal audio work, the 333A or 334A would be good choices. The auto-nulling will track slowly drifting oscillators or filters. I have occasionally been able to get a deeper null in manual mode (= to 331A and 332A in this mode).
Your Fluke 87 meters are very good and do tend to hold their calibration ... except you can not depend on the higher frequency AC readings at all. An HP 339A is more accurate on AC than a Fluke 87. Why you ask? Simple. The person who calibrates a Fluke 87 knows that once you change the case, or reinstall it, the AC readings take a wander. The most accurate way to align an 87 is to use another top cover with access points drilled out. Do the adjustments and reinstall the proper case. Then check to see how far out they are. I kid you not. If a replacement cover was not used, consider the higher frequency readings as if they are vague suggestions unless you have before and after readings in you hands that prove the readings are correct. I can say with certainty that one that is right on is a rare piece. Later Fluke meters that use what is called "closed case calibration" would be expected to be in tolerance.
Have I caught everything you were wondering about?
-Chris
anatech said:
Hi Chris,
Actually the calibration helped a lot. The analyzer was completely off, and couldnt measure anything. The readings were also completely off compared to the service manual: For ex, the voltage at A2TP8 was 2.6V when it should be 3.162. The amplitude pot also had to be adjusted as I was getting +0.3VDC when it should be -0.4VDC. Now, at least I can do some voltage measurements at 1-3V that were impossible before.
Regarding the trimmer cap's, dont worry, I just used a lot of DeOxit on them and now they look very shinny. Just kidding hehehehe, I know I shouldnt mess with them. I didnt even try to adjust them as they didnt need to.
I already 'cleaned' the controls with FOSPRO... something like DeOxit. But I just used a brush with some care. Maybre I should try the Q-tips??
Should I use Isopropyl alcohol? Or what?
Good to know I shouldnt use my Flukes at high freq.... mine certainly doesnt have any holes drilled for calibration!
Hi bsgd,
Normally, the meters on HP stuff are deadly accurate. Checking mine against an HP 34401A and an HP 3457A showed them to be dead on. I mean - dead on for all of them. That means that this instrument is actually going to be your primary asset when it comes to AC voltage measurements if it were only calibrated.
Since the adjustments have been altered, please find out how much it would cost to have the 339A calibrated at Agilent's nearest facility. Secondary choice would be a certified calibration lab, Fluke might be a good choice, after these, and other national calibration labs with traceable standards. "A guy" will not cut it here. After the calibration, your 339A will show an AC voltage to lower uncertainties than your 87 meters.
Digital is not always better, especially since they do not have "the other display". The one that shows you your measurement uncertainties. Dipping into the realm of metrology here, it is instructive to pick a reading as if it was displayed on your meter. Apply the measurement uncertainties (x% of range + basic uncertainty) and count error (at least + or - one count) and convert that to a voltage on the same scale. You will see a number that represents that the true value you measured could be the number you took down, + or - the calculated error you just figured out. For some of you with less expensive meters, you will realize that the last digit is often just for show. Some meters will even call the second digit to be a bit of a joke. The human animal likes to accept a string of numbers as "the answer" - exactly. Life is not like that in the real world.
For calibration uses, a 1% resistor is useless, 0.1% is realistic most times. Remember that you ideally would like a 4:1 TUR, or total uncertainty ratio. What that means is that your calibration standard (the correct reading with errors) should be 4 times better, or more accurate than the device you are calibrating. Put another way, the standard you are using should have at least 1/4 the uncertainty of the device you are calibrating - at the worst case. These resistors will be special order either way. Now, if you only had an accurate AC voltage source!
-Chris
Normally, the meters on HP stuff are deadly accurate. Checking mine against an HP 34401A and an HP 3457A showed them to be dead on. I mean - dead on for all of them. That means that this instrument is actually going to be your primary asset when it comes to AC voltage measurements if it were only calibrated.
Since the adjustments have been altered, please find out how much it would cost to have the 339A calibrated at Agilent's nearest facility. Secondary choice would be a certified calibration lab, Fluke might be a good choice, after these, and other national calibration labs with traceable standards. "A guy" will not cut it here. After the calibration, your 339A will show an AC voltage to lower uncertainties than your 87 meters.
Digital is not always better, especially since they do not have "the other display". The one that shows you your measurement uncertainties. Dipping into the realm of metrology here, it is instructive to pick a reading as if it was displayed on your meter. Apply the measurement uncertainties (x% of range + basic uncertainty) and count error (at least + or - one count) and convert that to a voltage on the same scale. You will see a number that represents that the true value you measured could be the number you took down, + or - the calculated error you just figured out. For some of you with less expensive meters, you will realize that the last digit is often just for show. Some meters will even call the second digit to be a bit of a joke. The human animal likes to accept a string of numbers as "the answer" - exactly. Life is not like that in the real world.
For calibration uses, a 1% resistor is useless, 0.1% is realistic most times. Remember that you ideally would like a 4:1 TUR, or total uncertainty ratio. What that means is that your calibration standard (the correct reading with errors) should be 4 times better, or more accurate than the device you are calibrating. Put another way, the standard you are using should have at least 1/4 the uncertainty of the device you are calibrating - at the worst case. These resistors will be special order either way. Now, if you only had an accurate AC voltage source!
-Chris
Hi bsgd,
Please check those test points and your regulated supplies for ripple. That could cause exactly what you measured (pit in stomach now). What you have reported is extremely abnormal.
Before you ever do anything, you must confirm that your power supplies are working correctly first.
-Chris
Oh-oh, not good!
Please check those test points and your regulated supplies for ripple. That could cause exactly what you measured (pit in stomach now). What you have reported is extremely abnormal.
Before you ever do anything, you must confirm that your power supplies are working correctly first.
-Chris
Hi Chris,
Im in Brazil, and things here are a little more expensive... I did find a place that could do a good calibration, but at $980 I dont want to.
I know a few 'guys' but I bet I can do better than them.
Thats why I'd rather try to fix it myself. If I need to buy something that doesnt cost a thousand bucks and that will be useful for me in the future to calibrate this thing, Im willing to.
But, I do think my calibration work was nicely done. I only did the gain calibration... and even after cleaning contacts and stuff... the measuring points of calibration are dead on... didnt change a bit. Since calibration is done at high voltages (by high I mean 1 to 3 Volts hehehe) and low frequency (that is 1kHz), I believe my Flukes (wich agree on all readings) did a nice work... I really think the problem lies elsewhere and is not a calibration problem anymore.
Im in Brazil, and things here are a little more expensive... I did find a place that could do a good calibration, but at $980 I dont want to.
I know a few 'guys' but I bet I can do better than them.
Thats why I'd rather try to fix it myself. If I need to buy something that doesnt cost a thousand bucks and that will be useful for me in the future to calibrate this thing, Im willing to.
But, I do think my calibration work was nicely done. I only did the gain calibration... and even after cleaning contacts and stuff... the measuring points of calibration are dead on... didnt change a bit. Since calibration is done at high voltages (by high I mean 1 to 3 Volts hehehe) and low frequency (that is 1kHz), I believe my Flukes (wich agree on all readings) did a nice work... I really think the problem lies elsewhere and is not a calibration problem anymore.
OK, I've got numbers!!
Here are my PS measurements.
Using my Fluke:
TP304= 14.94V
TP303= -15.08V
TP300= 15.19V
TP301= -15.20V
Using my PC Scope:
TP304= 15.6V
TP303= -14.8V
TP300= 16.3V
TP301= -14.4V
Now, the ripple numbers, all measure with my Scope:
Rpp= Ripple peak to peak.
ACrms= RMS AC Voltage present on the line.
TP304 Rpp= 40mV
ACrms= 6mV
TP303 Rpp= 43.8mV
ACrms= 8.10mV
TP300 Rpp= 50.1mV
ACrms= 5.8mV
TP301 Rpp= 46.9mV
ACrms= 6.10mV
I couldnt find in the manual the range this values should fall within. Where can I find this info?
Here are my PS measurements.
Using my Fluke:
TP304= 14.94V
TP303= -15.08V
TP300= 15.19V
TP301= -15.20V
Using my PC Scope:
TP304= 15.6V
TP303= -14.8V
TP300= 16.3V
TP301= -14.4V
Now, the ripple numbers, all measure with my Scope:
Rpp= Ripple peak to peak.
ACrms= RMS AC Voltage present on the line.
TP304 Rpp= 40mV
ACrms= 6mV
TP303 Rpp= 43.8mV
ACrms= 8.10mV
TP300 Rpp= 50.1mV
ACrms= 5.8mV
TP301 Rpp= 46.9mV
ACrms= 6.10mV
I couldnt find in the manual the range this values should fall within. Where can I find this info?
Hi bsgd,
That price is completely unreasonable. It's mostly an AC voltmeter! Surly there must be a competent place where you can get a proper calibration done. I didn't expect this to run you more than $200 on the outside. At $900, you could probably rent a Fluke 5520A multifunction calibrator for a week and make money on the deal! With that, you can calibrate most DVMs, oscilloscopes and all manner of AC voltmeters.
There must be some way you can have your instrument calibrated for reasonable money. You know what? Even an "in cal" HP 34401A plus a sine wave source would get you calibrated, a 3457A even. A newer Fluke might be enough to get the job done, but I'd be buying a nice bench meter with that money instead. I'm sure that with just some basic things, you would be able to set that meter up. In short, I completely agree with you. You can do this work.
I have not yet looked into my manual, and I would have to in order to give you a reasonable answer any further. I do know there are a couple wires near the capacitor that are sensitive to their location. Try not to move any of those wires.
If there is one under rated meter out there, it has to be the Agilent (HP) 34401A. It is extremely accurate, has a number of very useful functions and will average readings over a time period. If you were to buy anything, this one meter will make your life so much easier. If you can swing a new one, try the 34405A(?), 5 1/2 digit bench meter or go for the 34401A. It's still a current product! There are two improved models in the same class, then the 3458A laboratory meter. That one will not suit most audio work, and it costs more than I can wrap my head around.
I have to say that the one purchase I made at list that I never did regret was that HP 34401A. That one meter has saved me so much time, it's silly. I also use it as a calibration standard to set other meters up. It's still bang on with a freshly calibrated 3457A - not too shabby!
Is it at all possible that a "screwdriver jockey" got in there before you received it? I don't know it's history, but it is so unusual to see one way out like that.
-Chris
That price is completely unreasonable. It's mostly an AC voltmeter! Surly there must be a competent place where you can get a proper calibration done. I didn't expect this to run you more than $200 on the outside. At $900, you could probably rent a Fluke 5520A multifunction calibrator for a week and make money on the deal! With that, you can calibrate most DVMs, oscilloscopes and all manner of AC voltmeters.
There must be some way you can have your instrument calibrated for reasonable money. You know what? Even an "in cal" HP 34401A plus a sine wave source would get you calibrated, a 3457A even. A newer Fluke might be enough to get the job done, but I'd be buying a nice bench meter with that money instead. I'm sure that with just some basic things, you would be able to set that meter up. In short, I completely agree with you. You can do this work.
I have not yet looked into my manual, and I would have to in order to give you a reasonable answer any further. I do know there are a couple wires near the capacitor that are sensitive to their location. Try not to move any of those wires.
If there is one under rated meter out there, it has to be the Agilent (HP) 34401A. It is extremely accurate, has a number of very useful functions and will average readings over a time period. If you were to buy anything, this one meter will make your life so much easier. If you can swing a new one, try the 34405A(?), 5 1/2 digit bench meter or go for the 34401A. It's still a current product! There are two improved models in the same class, then the 3458A laboratory meter. That one will not suit most audio work, and it costs more than I can wrap my head around.
I have to say that the one purchase I made at list that I never did regret was that HP 34401A. That one meter has saved me so much time, it's silly. I also use it as a calibration standard to set other meters up. It's still bang on with a freshly calibrated 3457A - not too shabby!
Is it at all possible that a "screwdriver jockey" got in there before you received it? I don't know it's history, but it is so unusual to see one way out like that.
-Chris
All are flukes true RMScliffforrest said:
For use in electronics , must have digital multimeters with analog scale, the Fluke with this option is very expensive.cliffforrest said:
the accuracy of fluke is reasonable and better than competitors, you must be alert because which are "instruments of camp" (I do not know the word in English), means that the instrument is small and can carry, until such a machine for example, because the main market of the manufacturer is for Electrotechnics and electronics industry.
For measures of accuracy (damping factor, for example), you need a voltmeter for bench, the digital the Fluke manufactures, analog Leader and HP (see photo)
Well, My multimeter is a Protek 506 is it true RMS and analog scale and output RC232C (useful for monitoring the damping of the source in long time) cost...10 times less that the Fluke in the same category.
to measure the damping factor, I use the simulator
Attachments
My old Fluke 8020A definitely isn't true RMS.
Mind you, it's quoted accuracy is better than it's successor the 77, et-al.
It compares spot-on with my 87.
Superb meter, I use it all the time.
Mind you, it's quoted accuracy is better than it's successor the 77, et-al.
It compares spot-on with my 87.
Superb meter, I use it all the time.
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.