Forums :: Resources :: Features :: Photo Gallery :: Vintage Radio Shows :: Archives :: Books
Support This Site: Contributors :: Advertise


It is currently Jun Mon 17, 2019 2:01 am


All times are UTC [ DST ]





Post New Topic Post Reply  [ 19 posts ] 
Author Message
 Post subject: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150)
PostPosted: Jun Sat 16, 2012 3:52 am 
Moderator

Joined: Jan Thu 01, 1970 1:00 am
Posts: 2328
Location: Seattle, WA, USA
Fellow Forum Members,
I have succeeded in making reproduction HP 8601A output amplifier hybrids which match or exceed the performance of the original parts. For those curious about the process, read on! :D

First a bit of background. The story really starts with another project, reproducing the HP 1820-0285 ICs for the 8007 series pulse generators (see http://www.antiqueradios.com/forums/viewtopic.php?t=190955). After I finished that project I was really excited about the prospect of applying my skills to other obsolete / NLA ICs especially ones with broader appeal, as the 1820-0285 is really obscure. I have been interested in the possibility of reproducing the output amplifier for the 8601A ever since I read a post here on this forum asking for a replacement several years ago and thus, when I was contacted by fellow forum member bkharris about fixing a pair of HP 8601A sweep / signal generators, I gladly accepted.

After receiving the generators, I did some troubleshooting and discovered that the attenuator was blown in one, and it also had a bad output amplifier hybrid. The other one had a bad preamplifier hybrid but a good output attenuator and output hybrid, thus giving me a good set of hybrids for bkharris, and a pair of bad ones for me to try reverse-engineering. Furthermore, bkharris was extremely kind and donated the parts of the bad sweeper to me so I would have a testbed for the new hybrids. He also very kindly allowed me to hold on to both sweepers for about three months while I finished development of the hybrids (and finished repairing the good sweeper, although that is another story all together...) I owe him a HUGE THANKYOU for making this possible!

O.K. Now onto the actual project. For those of you who haven't yet gotten a chance to venture into an 8601A, the output hybrid is a little gold can as shown below. The one I reverse-engineered is marked 5086-7112, but the part was also sold under the part numbers 5086-3012 and 1820-0150.

Image

The top of the can is welded onto the bottom, so the first challenge was getting the darn thing open without destroying the very delicate circuity inside. I finally accomplished this by using a razor blade as a tiny saw to cut through the soft gold and copper of the seam in one corner. Once I penetrated the corner, I was able to work the razor blade along the seam and slowly separate the weld. Once I got it open, I used a microscope to create a mosaic of pictures and then I stitched these together into the composite shown below.

Image

As you can see, the device consists of a sapphire substrate with thin-film resistors deposited on it. (Thanks to K7KSW for correcting me on the substrate material.) On top of this, ceramic capacitors were soldered, and discrete bare-die silicon devices were attached and wire-bonded. Finally, the whole unit was adhered to the bottom of the can and wire bonded into place before the lid was welded on.

From the pictures, It was easy to determine that three of the devices were transistors and one was the detector diode, but the final silicon device proved a mystery. It looked like a diode, but measured the same conductivity in both directions, and since it was connected across the output resistor, it didn't make much sense for it to be a diode. After much simulation and experimentation I eventually determined that it was a MOS capacitor which serves to flatten the output by reducing the impedance of the output resistor at high frequencies and thus compensating for high frequency attenuation elsewhere in the circuit. Also note the wire bond jumpers which serve to adjust the values of what I am calling R3 and R8, and thus set the bias of the transistors.

Before proceeding the the actual reverse-engineering, out of curiosity, I used a high power microscope to examine the transistors, looking for the original failure. Sure enough, one of the transistors in the output stage showed an absolutely text-book electromigration failure. (Electromigration is a process by which metal gradually moves away from areas of very high current density to areas of lower current density, eventually causing failure by open circuit.) The picture below shows the failed transistor.

Image

I then proceeded to trace out the circuit, measure all the capacitor values (except the MOS capacitor) using an impedance analyzer, measure the values of all the resistors, and create the schematic below. This was easier then the 1820-0285 as the circuit is much simpler, but also more challenging in that, being an analog circuit, resistor values were much more critical. After discovering that resistors with the same dimensions measured radically differentially, I ended up calculating the values of all the resistors based on their W/L ratios and the resistivity of the thin-film material which I calculated from the value and dimensions of the output resistor, which I was pretty sure was 50 Ohms. I then compared the calculated values to the measured values and compromised appropriately, aided by an LTspice simulation. Also note that the schematic shows several resistors as parallel combinations of multiple resistors more on this later, but in the original each of these was implemented by a single resistor.

Image

Armed with this deceivingly simple schematic, I then set about designing a mechanical package which would fit in place of the original hybrid and laying out a board which to reimpliment the circuit. The package that I devised is shown later in this post, but the hardest part of designing it was finding appropriate pins. I ended up finding a Mill-Max part that works, but it was only available from Mill-Max themselves, not from any distributer. Luckily, though, I was able to get a sample, but once those run out I don't know what I am going to do. The main challenge with the board was that the collector resistors dissipate over a watt of power each! I couldn't actually fit 1W resistors in the tiny package, so I used multiple 0.5W 1210 package parts in series to implement the collector resistors. The other challenge was find appropriate transistors. They also needed to dissipate about a watt of power and have a transition frequency on the order of 1GHz. Fortunately, I found that BFG35 parts fit the bill. The board I designed at this point is shown below.

Image

After building it, the first step was to choose values for R3 and R8 (which were hand-tuned in the original) to get the right bias current. After a bit of experimentation I chose 2.2K resistors as this biased the transistors with slightly more current then my good working output hybrid, but less then the bad ones which I took apart, while keeping all the parts under their rated power specification. Unfortunately, upon powering it up in the 8601A, the output meter pegged and could not be reduced with the output vernier knob and there was, apparently, no output. It appeared as if the circuit was oscillating, as the detector was putting out several volts independent of input, but I couldn't see any oscillations on either my 400MHz oscilloscope or the 1200MHz spectrum analyzer. Very puzzled and annoyed that I had apparently wasted money on a run of boards, I put the project aside. When I got back to it, I had the realization that it could be oscillating above 1200MHz! Sure enough connecting it to the microwave counter revealed a ~2GHz oscillation! By hacking in a couple of emitter bypass capacitors I was able to quench it and discover that, otherwise, the circuit worked nicely!

Thus, with renewed spirit I designed a second board. This time, I was extremely careful to keep inductances low. I used large plains of metal rather then traces whenever possible and put the collector resistors in parallel instead of series. The result, shown below, is quite pretty, if I do say so myself. I also found 0.75W 1210 resistors which gave me a nice bit of safety margin.

Image

This revision worked on first power up, and after tweaking the output resistor bypass capacitor (C8) it actually proved to be flatter then the original, at least by my power meter! It also has about 3dB more gain then the original, easily meeting the 22dB specification. The picture below shows the completed product, both top and bottom, next to the original.

Image
Image

Testing of the open-loop flatness revealed, that aside from a small low-frequency bump, it was as good as the original. Results are shown below.

Image

Furthermore, testing the flatness with an RF detector reveals that, even under the fastest sweep rate, it is just as good as the original. Note that the droop at low frequency is due to my RF detector not being good below about 10MHz.

Image

Finally, I checked the harmonic content per the original manual and found it to be equal to or superior to the original across the entire band. Furthermore, there is less clipping at high output. Note that these units always clip at +20dBm into 50Ohms. From the original HP service manual "...noticeable distortion will almost always be present at some RF output frequencies with output levels above about +16dBm." However, my reproduction shows less distortion, and more symmetric distortion, then the original.

Image

Thus, I will call this project done! A zip file containing all of my design materials, including high resolution images of the original, an LTspice simulation, and the board layout files (to be used with ExpressPCB) can be downloaded at the link below. Also included is a service manual for the reproduction module including a parts list, schematic, and board layout diagram.

http://dasarodesigns.com/~medasaro/5086-7112/5086-7112.zip

Depending on the availability of the pins, and community interest, I plan to reproduce the preamplifier hybrid (which is also used as the loop amp) shortly. When that is done I will be able to put the bad sweeper that bkharris so kindly donated back together. Once again THANKS A MILLION to bkharris.

Here it is installed:
Image

As far as I know this hybrid was used only in the 8601A sweep / signal generator, but I would love to hear about it if anyone finds another application.

-Matthew D'Asaro

_________________
If it ain't broken, you’re not trying hard enough...


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Sat 16, 2012 5:53 am 
Member

Joined: Feb Thu 24, 2011 1:29 am
Posts: 3157
Location: Now outside Dallas, TX - with bobcats and coyotes and more! Oh my!
Matthew, you're pretty phly for a young guy. 8) What a fine job you've done.
A company called Susumu makes relatively high power SMT resistors (for their size). We use them in one of my denser analog designs.
I'm looking forward to the next adventure. :shock:

_________________
In a triode, no one can hear you screen.


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Sat 16, 2012 11:20 am 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 5287
Location: Montvale NJ, 07645
NICE!!

I may be needing one in my 8601 that is in the que to get fixed.


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Sat 16, 2012 6:31 pm 
New Member

Joined: May Fri 25, 2012 7:54 pm
Posts: 3
Location: Northern California
Matthew, very impressive replacement job. BTW, the substrate is sapphire (synthetic). That was chosen by Microwave division of HP because it has excellent RF/microwave characteristics. This was the first HP product to use sapphire thin film circuits. The 2nd product to use sapphire thin film circuits was the 8447A amplifier. An article on the HP 8601A appeared in the Aug 1968 issue of the HP Journal.

It was interesting to see the failure mechanism you found on the output amp.
Larry


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Sat 16, 2012 6:44 pm 
Member

Joined: Feb Sun 17, 2008 11:36 pm
Posts: 1995
Location: Dayton, Ohio
GREAT work Matt. I have two h/p 8601A's (both working, at the moment). I hope I never need to repair them. :shock:

Charlie


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Sun 17, 2012 12:56 am 
Member
User avatar

Joined: Oct Sun 15, 2006 12:57 pm
Posts: 6342
Location: Liberty, Missouri
Matthew,

Very nicely done. Nice to see unique projects like these. There are numerous HP hybrids that could use replacements.

_________________
Pics, Bold & underlined text, are usually links.

https://mikeharrison.smugmug.com/

Mike


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Sun 17, 2012 1:26 am 
Member

Joined: Jan Thu 01, 1970 1:00 am
Posts: 18525
Location: Albuquerque, NM 87123
How much are you asking for the output amplifiers? I have an 8601 which is working, and I want to keep it that way.


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Sun 17, 2012 3:11 pm 
Member

Joined: Jul Wed 22, 2009 3:07 pm
Posts: 761
Nice work Matthew, congratulation!

I did pick up an nonoperational 8601A last year. I was lucky, I only had to re-seat a board, and both hybrids in the output module. Now it works fine!

Peter


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Mon 18, 2012 5:08 am 
Member

Joined: Feb Sun 17, 2008 11:36 pm
Posts: 1995
Location: Dayton, Ohio
Johnnysan wrote:
How much are you asking for the output amplifiers? I have an 8601 which is working, and I want to keep it that way.


+1!

Charlie


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Mon 18, 2012 9:21 pm 
Member
User avatar

Joined: Feb Thu 08, 2007 12:36 am
Posts: 2139
Location: Hillsboro, Oregon
Beautiful work, Matthew. I've seen worse done by people who supposedly make a living at it.
Almost makes me wish my 8601A would... No! I didn't say that! (Hope it didn't hear me.) :D

I could not see the exact failure site in your electromigration photo. Can you point it out?

Thanks,
Dave Wise


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Tue 19, 2012 2:06 am 
Member

Joined: Jan Thu 01, 1970 1:00 am
Posts: 30698
Location: Pocasset, Cape Cod, MA
So that was you I talked to at MIT! We need name badges.


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Tue 19, 2012 3:07 am 
Moderator

Joined: Jan Thu 01, 1970 1:00 am
Posts: 2328
Location: Seattle, WA, USA
All -
Thanks for the outpouring of support! :D I really enjoyed doing this project and it sure feels good to hear that others like it as well.

Dave -
Quote:
I could not see the exact failure site in your electromigration photo. Can you point it out?


Thanks for the question. Here it is labelled.

Image

-Matthew

_________________
If it ain't broken, you’re not trying hard enough...


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Tue 19, 2012 5:28 pm 
Member

Joined: Jan Thu 01, 1970 1:00 am
Posts: 3808
Location: Westminster, CO, USA
Awesome Matthew. Really. The new board looks fantastic. Something to be proud of for sure.

Tony

_________________
Tony Casorso


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Tue 19, 2012 5:41 pm 
Member
User avatar

Joined: Feb Thu 08, 2007 12:36 am
Posts: 2139
Location: Hillsboro, Oregon
Thanks, Matthew, I get it now. So the symptom was gain loss due to loss of emitter area? That's subtle; I was looking for a short. I don't know what I was thinking. Great pictures, by the way.

Dave Wise


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Tue 19, 2012 5:50 pm 
Moderator

Joined: Jan Thu 01, 1970 1:00 am
Posts: 2328
Location: Seattle, WA, USA
Quote:
So the symptom was gain loss due to loss of emitter area?


Yes. Exactly, and the reduced gain of the transistor caused the bias point to change dramatically which caused clipping to occur as well.

-Matthew

_________________
If it ain't broken, you’re not trying hard enough...


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Wed 27, 2012 9:55 pm 
Moderator

Joined: Jan Thu 01, 1970 1:00 am
Posts: 2328
Location: Seattle, WA, USA
The issue I was having with the availability (or lack there of) of the Mill-Max 3169-0-00-15-00-00-03-0 pins has been solved. I found a vendor, Astrum Electronics, which would sell me a lot of 1000 pins at $0.375 each. Not exactly, cheap, but if I am going to keep making these things I need them so I went ahead and purchased them. Now that I have a source of pins, I plan to forge ahead with reproducing the loop amplifier / pre-amplifier module. The next problem to solve is how to open my failed module. Unlike the (newer) output module that I was able to force open with a razor blade, this one is very well welded. Grr.

If anyone out there has a broken preamplifier module, especially a newer one, I would love to hear from you. (The newer hybrids are identifiable because they have a blue and silver label with the part number rather then having the part number stamped onto the metal)

-Matthew

_________________
If it ain't broken, you’re not trying hard enough...


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Fri 29, 2012 1:28 am 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 5287
Location: Montvale NJ, 07645
Your work reproducing these unobtanium parts is super appreciated.


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jun Fri 29, 2012 8:03 pm 
Moderator

Joined: Jan Thu 01, 1970 1:00 am
Posts: 2328
Location: Seattle, WA, USA
Thanks! I managed to get my dead preamplifier hybrid open (ugly, but open...) so never-mind on my request for another one. The preamplifier hybrid is trivial compared to the output hybrid - it contains just two transistors and consists of a simple two-stage AC coupled common-emitter amplifier. I have it working in simulation, have the boards laid out, and have all the parts ordered so hopefully in a week or so I can release my design for this one.

-Matthew

_________________
If it ain't broken, you’re not trying hard enough...


Top
 Profile  
 
 Post subject: Re: HP 8601A Output Hybrids (5086-7112, 5086-3012, 1820-0150
PostPosted: Jan Sun 22, 2017 8:07 pm 
Member

Joined: Dec Wed 24, 2014 7:34 pm
Posts: 2009
Location: Weimar, Texas
This thread is only 4.5 years old but it seems like a good place to say

The hybrid module was very easy to install and it "works a treat" as a certain YouTube guy would say. Well done Matthew.

_________________
It's hard to solve an equation if every term is an unknown.

73
NE5U

Mike


Top
 Profile  
 
Post New Topic Post Reply  [ 19 posts ] 

All times are UTC [ DST ]


Who is online

Users browsing this forum: Beaconhunter, bob91343 and 10 guests



Search for:
Jump to:  




























Privacy Policy :: Powered by phpBB