Home -> Testing -> HP8640
  Bench   HP8640

HP-8640 Service Page

Note: None of the contributors or myself assume any liablility for any damage caused by acting on information contained herein (of course).



SITE UPDATES

For additional information regarding repairs to the HP8640 series generator see "Hints and Kinks":

Most recent HINTS and KINKS Update, November 28, 2007 see "No. 14."


The HP8640 signal generator, though out of production for many years, is still a real jewel when it comes to HF/VHF signal generators. It is a very rugged instrument with a wide frequency range (0.5 MHz to 500Mhz), with RF output depending upon model number between 16 dBm to over 20 dBm and a built in attenuator that attenuates the RF down to -140 dBm accurate to within + or - 0.5 db. I obtained a used HP8640B in 2002 and it worked flawlessly until one day the output became intermittent. This led me down the path of looking for the problem and finding a solution. The problem was the pre-amp. hybrid amplifier. Since this part is no longer in production I decided to try and replace it with a three terminal power amplifier, the MAV-11. The solution was described in QEX July/August.

Here is an abbreviated version of the article:

Help For The HP8640

By Markus Hansen, VE7CA

If you own a HP8640 signal generator and the output power has become intermittent or it has quit working all together, then read on.

1. Locating The Problem

On the back of the generator is a large heat sink and extending from the heat sink is a BNC connector marked Aux. Output. Connect a scope or a RF power meter to the BNC connector and turn the generator on. If there is no output at this point, the problem is either the hybrid pre-amp., the supply voltage for the pre-amp. or any of the circuits preceding the pre-amp. If there is no RF at the BNC outlet, turn the generator up side down and remove the bottom cover. Turn the generator so the back is facing you and remove the cover from the AM/AGC AMPLIFIER ASSY which is located in the lower right hand corner. This assembly is divided into two sections, the A26A3 Modulator Assy. is on the left side (which contains the pre-amp. hybrid) and the A26A1 Power Amplifier and AGC Detector Assy. on the right side.

In each section you will see what looks like a large power transistor in a case style similar to a TO-3 or TO-8 transistor. The hybrid amplifiers are contained under the caps of these units. On the PC board, in the left section, locate the end of the 50 ohm coax line with the center connector terminating in a circuit pad. (see Fig. 1)

Fig 1. A26A3 Modulator Assy



Turn the frequency range switch of the generator to its lowest range and turn the generator on. With a scope determine if there is any RF at the end of the 50 ohm coax. If there is, and it is a square wave measuring 550 mVp-p, then you have confirmed that everything before the pre-amp is working and the problem is either the pre-amp or the power supply. The only way to really find out if it is one or the other is to remove the pre-amp. hybrid by removing the two screws holding it in place. Once the pre-amp. is removed check the supply voltage that it is +44.6 volts. The +44.6 volt input is clearly marked on the PC. If there is no voltage present or it is the wrong value then proceed to fix the power supply problem. The reason that you have to remove the pre-amp. to check the voltage is that the pre-amp. may have an internal short, which was the case with my pre-amp. The power supply has what is called a crow-bar shut down feature, where if the output is shorted, the voltage goes to zero preventing damage to the power supply.

2. Changing Supply Voltage For The MAV-11

The MAV-11 requires 5.5 Volts while the hybrid amplifiers both run on 44.6 Volts. The 44.6 V line enters the AM/AMPLIFIER ASSY box that contains both amplifiers via a multiple pin connector and then along a PC board trace in the center of the box. There is no spare pin on the connector so a feedthrough capacitor has to be added to bring in a voltage line for the MAV-11.

With the back of the generator towards you, take off the top cover. The AM/AGC AMPLIFIER ASSY is now located in the bottom left corner. Remove the AM/AMPLIFIER ASSY from the generator by following the instructions on the amplifier lid. Once removed, turn it so that the side opposite the hint sink is facing up. There you will find a removable rectangular aluminum plate. After you remove the screws and the plate you will see a row of feedthrough capacitors in the cavity below. From one end there are two wires that go from feedthrough capacitors to the section that contains the hybrid pre-amp. The second wire (white with purple and brown bands) from the end is connected between the feedthrough capacitor +44.6 volt input line and a pad on the pre-amp. PC board that is marked 44.6 V. Unsolder this wire from the feedthrough capacitor. Make a mark on the aluminum plate in the middle between the two end feedthrough capacitors. Drill a hole to fit a feedthrough capacitor. I used a good quality 0.001 uF feedthrough capacitor that I had purchased from Down East Microwave. One end is threaded so that when it is screwed into the aluminum plate the result is a very secure RF connection. Connect the wire that was soldered to the +44.6 V feedthrough insulator to the new feedthrough insulator using heat shrink tubing to cover the solder connection. Connect a 12" length of insulated wire to the outside of the new feedthrough insulator and again pass short length of shrink tubing over the connection. See Fig 2. Now reassemble the AM/AMPLIFIER ASSY and insert it back into the generator.

FIG 2.



The other end of the wire just attached to the outside of the new feedthrough insulator is now soldered to a hole at the end of a trace that leads to C4 that is on the power supply rectifier assembly circuit board. I choose this point as the voltage is approximately +15 volts leaving sufficient room to drop the voltage to 5.5 volts for MAV-11 while limiting the current to 60 ma. Also this particular winding has the largest current capability of all the transformer windings. C4 is 8200 uF and large enough that I was not able to detect any ripple on my scope.

3. Substituting A MAV-11 For The Hybrid Pre-Amp

I had some Mini-Circuits MAV-11 amps. on hand so I decided to try one as a substitute for the pre-amp hybrid. You may want to choose another model such as the MAV-11SM or ERA-1SM that has a typical gain flatness up to 2 GHz of +/-0.5 dB., 0.02 dB better than the MAV-11, though I don't think it will make much difference because the generator has an AGC circuit to keep the output power at a constant output.

Begin by removing the cap from the existing pre-amp. module. I inserted the module in a vise and using a metal punch I was able to get under the seal of the cap and pry it up and off. I removed the hybrid from the base by tapping the edge with a small screwdriver. I decided to use the original base of the hybrid because it has the connecting posts that connect through to the PC board below and it has room on top for a MAV-11 and associated parts. I cut a small piece of PC board to fit between the connecting posts and extending over the mounting holes of the base. If you use a MAV-11 then drill a hole in the center of the PC board large enough so that the MAV-11 drops into place and the four leads lie on top of the PC board. I used a Dremel tool to cut isolation pads for the input and output pins of the MAV-11.

Facing the back of the generator and looking at the base that holds the hybrid pre-amp., the connecting posts farthest away from the back of the generator from left to right are: output, ground and input. The connecting posts closest to the back are the left post (not used) and the right post, (voltage supply input). From the base input post connect a 4 dB pad in series with a 0.1uF monolithic or surface mount capacitor. The other end of the capacitor goes to the input pin of the MAV-11. Between the voltage supply input post on the base and the output pin of the MAV-11 connect three 470 ohm 1/4 w resistors in parallel. The three parallel resistors should measure about 157 ohms and serve to drop the supply voltage from 15 volts to approx. 5.5 volts and limit the current to 60 ma. From the output pin of the MAV-11 connect another 0.1 uF monolithic or surface mount capacitor to the output post of the base. See Fig 3. for the schematic.

FIG 3. Schematic



Before putting the AM/AGC AMPLIFIER ASSY cover back on, make sure that the new components do not touch the lid of the cover. I cut a small single sided PC board to cover the MAV-11 and new components and soldered the copper side of the PC board to the screen that is sandwiched between the cover and the assembly box. It took me several hours to perform the above changes. I strongly recommend that you do not hurry when making the suggested changes. You don't want to make a mistake that may cause other problems that are not fixable because of the scarcity of some of the HP original parts.

4. Summary

When I completed the substitution and turned on the generator, with the attenuator set to 0 dBm, the generator output meter read 0 dBm and my RF power meter connected to the output connector also registered 0 dBm. This was with the frequency set to 875 kHz. I was one very happy person. According to my RF power meter the output is within +/- 0.2 dB across the generators full operating frequency range. (I had previously calibrated the RF Power Meter using the signal generator as the source and confirmed the calibration using another HP8640 that a friend owns.) The repaired generator also produces full power across the full frequency range of the generator, which for the HP8640 Avionic Version is +13dBm. The AM and FM modulation features of the generator were not affected by the substitution.

FIG 4. HP8640 with Power Meter



Obviously, the problem outlined in this article is only one of many reasons why a HP8640 series signal generator may not be working. However, it became clear after discussions with a number of fellows hams that had years of experience working with this particular series of generators that the hybrid pre-amp is prone to failure. Hopefully this article will help those who experience a similar problem. It is interesting to note that there is a notation on the circuit diagram for the Avionic Version of the HP8640 indicating that the output amplifier is a 16 dBm amplifier. The MAV-11 may therefore also work as a substitute for the final amplifier hybrid. This would not be the case for other versions of the HP8640 as they typically have a higher output power than the Avionic Version that has a max. output power of 13 dBm. The MAV-11 is listed as having a max. output power of 18 dBm (1 db Comp. Typ). If it was used to replace the final power amplifier in the higher power generators, then the AGC circuit will not be effective unless it is adjusted to recognize that the new amplifier is only capable of putting out 18 dBm max. There may be three terminal amplifiers available that provide more than 18 dBm that would be a better substitute for the final amplifiers in the higher output generators.

I have not been able to confirm this, but I was told that the last year that the HP hybrids were being produced they sold for over $200 US. I paid $2.65 US for the MAV-11's when I purchased them a couple of years ago. Now that is progress!



HP8640 Hints and Kinks:

If you have a solution for fixing the HP8640, especially if it involves repairing or finding substitutes for HP parts that are no longer in production please let us know and we will post it on this web page.

HP Part Subsitution:

Frank Carcia, WA1GFZ shared a list of HP part number conversions to industry-standard (JEDEC) parts. Click
here to see the list.


1. Output dies on tuning to high-end on all bands,   November 2004: Markus, VE7CA

This may be due to a problem with the 128-256 MHz band ECL divider in RF Divider A10A2, the board located just under the large die-cast cover on top-side of generator. If there is input at the divider over the full freq. range but the output fails above a certain freq. then this fix may apply.

A factory modification was made: 100 ohm variable resister R6 was added from pin 13 of this divider to ground. This mod. increases the sensitivity of the divider. To adjust the resister, tune gen. to high frequency end of band and set resistor half-way between the high and low cutout. If the bias is too high or low, the divider will not clock.


2. Cracked Wheels,   November 2004: Hadley, K7MLR

I just finished the first read of your article, quality work, thank you very much. There are two other, very common, problems with the HP8640:

  1. The nylon gears, in the gear trains, of the audio osc. and range switch crack and will eventually fall apart. If caught in time, epoxy will extend their life.

  2. The small finger contacts, mounted to the clear Styrene wheels, located on the range switch assembly and the attenuator will become intermittent. Again with some work and epoxy this is pretty easy to fix. The Styrene wheels are reversible and give new purchase points to re-mount the fingers when they fall off. Every HP8640 I work on gets the epoxy treatment on the wheels because the fault will occur sometime down the road. Just a warning, don't lose any of the fingers during disassembly or you will be up the creek unless you have a hanger queen.
Ed. Note: Picture of a Cracked Styrene Wheel on EB5ABV's Web site is here. It shows a cracked wheel before repair and the small finger contacts, mounted to a clear styrene wheel.


3. More Cracked Wheel Help,   November 2004: Ken VE3FIT ve3fit@rac.ca

I read your article in QEX about repairing an HP8640 and it led me to your web site. I greatly enjoyed both. Last winter, I restored an 8640B which I had bought "as-is" on e-Bay. Fortunately, it was a long winter: I did a repair on several cracked Delrin gears and in a nutshell, here it is:
  • Having removed the gear, I put it in a refrigerator's freezer compartment for a few hours. This will cause the delrin to shrink more than the brass hub. Drift out the brass hub. Mine came out with just a gentle tap on a hammer and drift. (Credit the next part to Jeff Liebermann). Use 24 hr epoxy to glue the Delrin back together. Use a round file to open up the delrin centre hole so that the brass hub will just fit back in. Epoxy the hub back in.
I had to replace the final power amp in my unit. Fortunately, I managed to get a 'pull' from Fair Radio, but it cost about US $125. I was just looking on the WJ Communications web site and it looks like they now have several MMICs that could work in the 8640B. Check out their ECG008.

It's great that you have offered your web site for posting information regarding the HP8640.


4. Caution when storing HP8640,   November 2004: Terry K7TAU

A retired co-worker (George Steen) has located a fellow that worked at HP on the HP-8640-B RF generator and has been corresponding with him regarding the instrument. Here is something George passed along to me and suggested that I copy you on this information.

"Word of caution: don't store the oscillators on-end (with their tuning shaft pointing up), because if you do so, eventually the internal silicone grease migrates down to (and saturates) the back plate where the transistor oscillator resides. The grease sometimes gets between the plate and the case, causing loss of good electrical connection. End result: the unit drops out of oscillation at parts of the band - - or in bad cases, won't oscillate at any part of the band. " This is a direct quote taken from correspondence between George Steen and Greg Burnett, Retiree of HP.

Thought you might be interested in reading this and perhaps posting this on your web page.

Regards, Terry K7TAU


5. Using a surface mount type MAV-11 and PCB,   December 2004: I5SXN Angiolo Chiti

I am I5SXN, an old Italian radio amateur no more active but I am now only a technician. In 1999 I had the same problems on two generators HP 8640B. At that time HP asked me $600 for that modules. I solved the problems in the same way, as you, but I opened the container of the amplifier and I introduced inside a small printed circuit with an MMIC type MAV11, two condenser and a lot of chip resistors series and parallel to obtain the right current for the MMIC and dissipation using directly the 44 volt supply. I closed the container and in this way if you open my HP8640 you do not see any transformation. I send You the schematic diagram and the printed circuit indications.

PC Board Layout & Schematic



Printed circuit and mounting position of the components. The shape is octagonal for problems with my CAD, I rounded the circuit before mounting in the container. Diameter 25 mm. Distance between pins A and B 12.5 mm, and between B and C 16.4 mm. The circuit was of 0.8 mm thickness, with copper on the lower side. I scratched the copper around the active pins. With small holes and wires I connected the ground pins of the MMIC to the lower copper side. I used the original pins of the module. Another time I repaired the burned final amplifier adding a zero bias diode on a different printed circuit. I hope to be useful to you. Good luck and 73.

Angiolo Chiti I5SXN, E-mail: a.chiti@rossbauer.it


6. Another replacement type for the MAV-11,   May 2005: Tom Bruhns K7ITM

I just photocopied a mixer article from QEX (July/Aug 2004 issue) and happened to get the last page of an article you apparently wrote about HP8640 amplifier issues. In footnote 4, you mentioned a WJ amplifier. For a little better output power and quite a bit better IP3, have a look at Sirenza SBF-5089. It's rated nominally for 500MHz high end, but the rolloff is gradual and it should be no problem using it in the 512MHz HP8640B. It's probably a bit shy on power output to be used directly in the output stage if you need full power output. The SBF-5089 has a particularly high IP3 (which is why I know about it...), and you probably don't need that linearity in the 8640 output. For power output from a monolythic gain block, have a look instead at the Sirenza SGA-7489, which should give you a bit better IP3 than the WJ part and about twice the output power--maybe enough to actually give full 8640 output. Plus, it's rated for a wider frequency range.

Cheers, Tom Bruhns


7. Help for Oscillator Failure in the HP8640B.

John Klingelhoeffer, WB4LNM wrote an excellant article in the Sep/Oct 2005 issue of QEX describing how he fixed the oscillator in his HP8640B generator. John describes in great detail the process of disassembling the A3 oscillator assembly in order to get at the oscillator circuit. He included many clear pictures of the inside of the generator and oscillator assembly. If you don't have a copy of the QEX article you can order a copy from ARRL.


8. Oscillator Frequency Adjustment HP8640B.

If you set your HP8640B so that the frequency counter reads exactly 10.000 Mhz but the output frequency is not exactly 10.000 Mhz, you can easily calibrate it yourself.

  1. Remove trim strip that holds front panel window in place. Gently pull window up and out and remove it. Allow generator to warm up for 2 hours.

  2. While listening to WWV at say 10 MHz in a AM receiver,(with a a short length of wire attached to the generator output), if your generator is off frequency even though the frequency counter reads 10.000 Mhz, then you will hear a beat note over-riding the WWV carrier. Adjust the time base adjustment pot (available through the hole in the front of the counter casting) until the oscillator is zero beat with WWV at 10 MHz, i.e. the beat note disappears.

  3. Replace front panel window and trim strip.


9. More Gear Information for the HP8640B.

In order to purchase new plastic gears for the HP 8640B one must specify the Diametral pitch (DP), the Pressure Angle, and of course, number of teeth, bore, thickness, etc. The DP is 48. A new 20 deg. pressure gear does NOT mesh correctly with the original gears. Hence, I conclude the pressure angle is 14.5 deg., the other choice. I have not tried a new 14.5 gear with the originals, however. I used Super Glue to mend split gears and it worked well. Clamp the pieces together for a few hours. I could not pull the joint apart with my hands.

NOTE: The latest available manual from 1992 for the 8640-B surely covers the "new" assembly and it gives a detailed gear specification for the combination gear:
"48-T 48-DP 14.5 DEG-PA".<
Regards, and tnx for your help. Jim, K8JL.

NOTE: The latest available manual from 1992 for the 8640-B surely covers the "new" assembly and it gives a detailed gear specification for the combination gear:
"48-T 48-DP 14.5 DEG-PA.


10. More Information regarding Storing the HP8640.

Hi Markus:

I E-mailed you awhile back about HP8640B problems. I stored my 8640 face up. When I read your webpages, I saw the comment bout intermittent frequency operation. Sure enough, I had 7 dead sections in the tuning range, including both ends. Additionally, the generator would not lock, even after an hour warmup. I E-mailed Terry, K7TAU, for help. He recommended operating the unit face down, to let the silicone grease flow back. After three months in this position,the HP8640 had only one dead spot in the middle of the range and would lock. I turned the unit off and let itset flat for another couple of months. I turned it on before Xmas, and lo and behold, the oscillator works perfectly. The bottom line, you don't have o open the unit and dissemble and clean the cavity oscillator to restore proper operation. You can let it sit face down, but it will take some time to come back to full operation.

Thanks for your help, and your webpages.

Jim Korenz, N8PXW


11. Information regarding variable and fixed audio irregularities in the HP8640.

After finding and reading through the HP repair information on your site, I thought the following may be of some interest to fellow HP user's. My HP8640A developed a fault causing the variable and fixed audio oscillator to stop working. I located the problem to the thermistor module and discovered that the fault seemed to be an open circuit between the common point (A) and the two series connected thermistors to connection point (C). The path between (B) to (A) was ok. No information is forthcoming in the HP manual regarding this component.

After trying several sources for a replacement and in most cases only being offered a complete replacement board for £50 plus vat etc, decided a cheaper cure must be sought. Trying various remedies I recalled that a solution might be to use series connected lamps to replace the whole original thermistor sealed assembly. I happened to have a supply of miniture 16v lamps so connected 3 in series and placed them on the top of the pcb soldering directly to the pads and interconnecting with thin wire.

The idea worked OK apart from a slight bouncing of the o/p amplitude at switch on. The output remains stable otherwise. It may not be the best of "fixes" but it cost nothing to achieve and will certainly suffice for my needs.

Regards, Steve GW4ZDU


12. Information regarding manuals for the HP8640.

Aglient Technologies who purchased HP has now posted operating and service manuals for many of the older HP test equipment.

Click here to go direct to the HP8640 service manual site:
HP8640 Manual.


13. October 2007 information regarding Combination Gear Replacements.

One of the problems that many owners of the HP8640 series signal generators face is finding a replacement for the combination gears (( "Fig. 8-97, A9 Peak Deviation and Rang Switch Assy" it is called: Item Number 13 and 19 (they are identical!), Reference Designator A9MP12 and MP13, Description: Combination gear. )) These gears have a tendency to split with use and if not replaced they eventually fall apart. Using epoxy, some have been successful in repairing the gears. he attached picture shows the break in the gears in my old HP8640.

Fig 5. Split Gear in Combination Gear Assembly



Recently I recieved an email from Mr. Wieschhoff in France regarding the Split Gear Combinatin:

"After having followed the conversation of some interested circles in early 2007 about the re-manufacture of the combination gears in the HP generator 8640B I tried to get some information about the follow-up and result. The early generators manufactured containing what I would call the "old" dev./freq. switch assembly. The difference to the "new" one is that it uses standard type switch segments where as the "new" ones uses a printed circuit type rotary whiper discs to switch. Apparently both assemblies were manufactuerd in parallel for some time, since the manual from 1976 covering the generator version HP8640-A shows still the "old" assy. The latest available manual from 1992 for the 8640-B surely covers the "new" assembly and for the fist time I see it gives a detailed gear specification for the combination gear: "48-T 48-DP 14.5 DEG-PA". It may also have been given in manuals between 1976 and 1992 but I do not have any. In ALL manuals quoted (including the one covering the "old" version) the part numbers for these gears are identical: 1430-0773 and 1430-0774. In fact both different numbers describe identical parts. Why HP did this I do no know but it may have been for internal procedural reasons. I used a 0734 replacement gear bought from HP some years ago (unfortunately at the time I only bought one) in place shown for the 0774: no problems."

Best regards, Reinhart


14. November 2007 Final Hybrid Amplifier Replair.

I recently made a small SMD board to plug in in place of a blown output hybrid in a 8640B. I used the Philips BGA6589 and could reach about +6dBm. Whilst this got the machine back in working order it was not as linear when modulated really only got to 0dBm.

I started to wonder why the hybrid had died since it showed output on the meter but none at the output pin. I concluded the either the output C had become detached or it’s bond wire to the output pin broken. So I mounted the hybrid on a hardwood mandrel and centred it in the lathe and gently turned off the top.

Fig 6. Final Hybrid Amp. with lid removed.





As the picture shows the bond wire to the output was vaporised! The output C had a slight coating of gold sputter but was otherwise OK. A small semiconductor pad capacitor which had been bonded to earth had also melted and was undoubtedly the path to earth for the blast of volts which took out the hybrid.

I soldered on to the output pin an 0805 50V 0.1 and connected it to the original output C with a single strand of 0.2mm. The added capacitor is because the old output C has now 10K of leakage. I re-inserted it in the 8640B put a piece of cling film over it and all is back to normal.

Click Here to view a copy of the circuit for the final Hybrid Amplifier.

" The schematic attached is an educated guess at the hybrid by looking closely at the blown up photo. The resistor values are scaled by measuring the square area of a couple of known resistor values and applying this to the other resistor tracks. Capacitors are estimated. The inductors are calculated. This excellent job was done by Richard, G3SHK who made this sort of stuff in a past-life at Philips labs".

I am pretty certain a good number of old output hybrid may have been zapped by working on valve kit as the DC input voltage of the spec is very low. If anyone has hung onto the hybrid I recommend peering inside! Mine had a sapphire substrate but since you might encounter Beryllium Oxide turn at low speed with care and wear a face mask!

73 Robin G3LBA


If you have any solutions for problems you have encountered for the HP8640 series signal generators, please let us know at: