HP8640B-opt323: Should I return it?

[tatel]

This military version of HP8640B is my first piece of RF equipment, not counting a dummy load.

Matter is:

a) FM deviation knob jumps over some detents going CW, then knob marks do not coincide with detents. Going CCW, it also jumps some detent, but not so much, and knob marks again coincide with detents. I'm afraid this could mean gears are bad. See it here: https://youtu.be/Zajaf_QCRI0

b) Output level is lower when the knob is on 20 dBm mark than on 10 dBm mark (meter maxes out here). Vernier knob is fully CW in both cases. WTF? See it here: https://youtu.be/GjVMWE5l6xA

So, what the old wise men think about it? Should I return it?

[tatel]

Vendor says it was in "perfect working condition" when shipped. 

Please enjoy attached pictures.

[Stray Electron]

  If you bought it from E-bay then you're entitled to return it and to get absolutely full refund, including the shipping costs, if it was categorized as working. Even if it was described as "As-Is".  Ebay and PayPal are both quite clear about that.

   That said, the 8640s are quite old and are well known for having these types of problems so if you really want one then you might have to grit your teeth and keep this one and try to fix it. I'm very skeptical about the chances of finding any 8640 that doesn't have the same problem.  But depending on how much you paid for it, you're probably entitled to at least a partial refund.  Unfortunately that is up to the seller and if he refuses to give you a partial refund then you're either going to have to return it for a full refund or keep it and get nothing.

   I can't tell how many gears in yours are cracked but as of a few years ago there were NO replacements for those so if it were me I would return it and give up on buying an 8640.  I passed on one in a surplus store a few years ago for $25 because I could tell that it had similar problems and, to me, it wasn't worth the effort that it would require to fix it. Even if it was possible. 

[ZigmundRat]

To return it or not I guess depends on what you want to do. But these units (I had one for years) are notorious for broken gears. In fact it would be surprising to find one today with them intact. These are commonly repaired with a tie wrap around the body to hold them together. But you can also get replacements - there’s a fellow in India selling brass ones on eBay. Either solution is a fairly straightforward mechanical repair.

The output control is another matter. It sets the max for that range. If the vernier is fully CCW, it should not pin the meter. It should be variable from something below 10dB down from max, up to full scale.  It sounds like you may have a problem with the attenuator, in which case I would return it. They are usually not repairable. (At least easily.)

[tatel]

  If you bought it from E-bay then you're entitled to return it and to get absolutely full refund, including the shipping costs, if it was categorized as working. Even if it was described as "As-Is".  Ebay and PayPal are both quite clear about that.

Yep. Unfortunately no eBay nor Paypal used. This is a surplus seller. So he has a reputation and a business to take care of. Since we are still negotiating, all I can say is: if there is no acceptable solution by Wednesday, you will know which vendor it is.

[tatel]

To return it or not I guess depends on what you want to do. But these units (I had one for years) are notorious for broken gears. In fact it would be surprising to find one today with them intact. These are commonly repaired with a tie wrap around the body to hold them together. But you can also get replacements - there’s a fellow in India selling brass ones on eBay. Either solution is a fairly straightforward mechanical repair.

The output control is another matter. It sets the max for that range. If the vernier is fully CCW, it should not pin the meter. It should be variable from something below 10dB down from max, up to full scale.  It sounds like you may have a problem with the attenuator, in which case I would return it. They are usually not repairable. (At least easily.)

Yep. Real problem is output level. Vernier works, but meter maxing out in the 10 dBm range and getting lower on 20 dBm range is worrisome. Warning light for excess output power never turns on.

Should gears be the only problem, a partial refund would be acceptable. Asking for a replacement. He said he has more 8640s. I guess he doesn't have any with good working gears. A replacement device with good output levels would be acceptable, with a partial refund if gears are bad. He got about €350 saying it was in good working condition, so that partial refund would be in order.

But he has to agree, or be hit on his reputation. We will see.

[tatel]

Well, a complete refund has been offered. Seller will also pay return expenses, so he is OK-ish.

Now, before effectively returning it, I would like to check output levels. If all the problem is bad gears, device would remain here. I see contacts on this military version are riveted into some kind of board. I like it, puts out one of the most mentioned source of problems on this device. Please see first attachment.

From civilian version manual, it seems that output level should be read on the meter, on two different scales (0-3 and 0-10) and also different units. Both the scale to read from and units to be employed would depend on the range selected on the knob. If max output level is +19 dBm/2Vrms, that would explain why the meter goes to 2 in the 0-3 scale when the knob is in the 20 dBm/3Vrms range. Then perhaps it should go to the top of the 0-10 scale when the knob is on the 10 dBm/1Vrms range. Which is what happens, please see videos linked on OP. Am I getting this right?

Alas, the procedures mentioned in the civilian version manual to check output levels require quite a bunch of equipment that is not here. All I have at this moment is a 100 MHz BW oscilloscope. N-to-BNC adapter is on the way. I'm reading output levels with oscilloscope probe on the pin of N connector. TinySA + some attenuator could be purchased. Any suggestions would be greatly apreciated.

Frequency readings on 8640B's counter and oscilloscope almost exactly agree up to where scope is able to reach (about 325 MHz). Sine wave looks good except in the 32-64 MHz where I see it distorted. Please see second attachment. Any thougts?

[tatel]

Look as if there is some problem with second attachment...

EDIT: Bad measurement procedure, this is probably a reflection. Distortion vanished with a more correct measurement procedure.

[tatel]

One last thing: I was expecting output level be the same on all the frequencies, as long as output level knob remains in the same position. However output level varies with frequency. Is that the usual behaviour?

[tatel]

Attachments are getting messed up. Here it is again the riveted contacts one

[Ice-Tea]

One last thing: I was expecting output level be the same on all the frequencies, as long as output level knob remains in the same position. However output level varies with frequency. Is that the usual behaviour?

Depends on how much it varies. If it varies by 1-2dBm then that is to be expected, especially for an old, uncalibrated and unadjusted device.

[tatel]

EDIT: I messed seriusly these measurements.

I would say it varies quite a bit more than 1-2 dBm.

As said, I'm measuring with oscilloscope probe on N connector pin, short ground spring to N connector sleeve, 50 ohm pass-through terminator on oscilloscope input. Frequencies are in all cases in the middle of the range. Last one could be attenuated since it's near oscilloscope's -3 dB limit. These are the results:

.5   -    1 MHz ---------------------->      52 mVrms
1    -    2   "    ---------------------->    102     "
2    -    4   "    ---------------------->    200     "
4    -    8   "    ---------------------->    368     "
8    -  16   "    ---------------------->    616     "
16  -  32   "    ---------------------->    906     "
32  -  64   "    ----------------------> 1.26    Vrms
64  - 128  "    ----------------------> 1.21       "     

Maybe it could be adjusted. We have an organization of pirate FM stations and there could be a power meter sleeping somewhere. Or perhaps this is a red flag and device should be returned ipso facto.

[Ice-Tea]

What do you mean 50ohm termination on scope input?

[tatel]

Oops... your comment made me realize that I seriously messed up my measurements, not just because I used terminators, but because probes were on 10x attenuation. Sure enough, with probes on 1x, no termination, I'm getting about 2.25-2.35Vrms on all ranges (I mean the lower ones, up to 20 MHz which is the BW of my best probe on 1x).

I expect the N-to-BNC adapter will be here this week. Then I will measure all ranges up to 100 MHz with coax cable. I guess I'll be unable to understand the meter until then.

What a wonderfully helping resource this forum is. Thank you very much.

EDIT: Distortion on the 32-64 MHz range is now missing. I suspected I was having some reflection but it should have made me thinking again about my "measurements"

[tatel]

Adapter arrived. So I measured output levels again. N-BNC adapter, 50 ohm coax, 50 ohm terminator, oscilloscope.

Output levels seem to be mostly right. Yeah, generator meter has to be read on the 0-3 and 0-10 scales, depending on the position of the output knob. If it has some kind of "3" mark on the inner circle, 0-3 scale has to be used. If it's a "1" mark, 0-10 scale is used instead. No fancy light indicators.

Output is basically the same on all frequency ranges as long as output knob remains untouched. On 48 MHz and above I see generator meter maxes out on the 1 volt output level range. However I read lower and lower values. Surely enough the reason is oscilloscope bandwidth.

Lectures are flat up to 24 MHz, then at 49 MHz and above they get lower and lower. Oscilloscope (GW Instek GDS1054B, upgraded) -3 dB limit is at 109 MHz.  It nicely coincides with I was expecting form risetime (about 3.2ns) measured with Leo Bodnar's pulser.