Stupid JFET, WJ-8888

[KA3EKH]

My Watkins Johnson 8888 uses some weird form of N-Jfet in its front end and I need a new one. The marking on it is something stupid like U320, I know that the WJ-8718 and other WJ junk uses CP643 N-Jfet but do not know if there is a generic low noise fet that I can stuff in the radio. Also have to remember, do not leave receiver connected to antenna when transmitting! And if anyone happens to have a schematic for a WJ-8888 that would be very helpful
Ray F. KA3EKH

[Pete, WA2CWA]

Data sheet attached

[N4LTA]

The U310 fets are still available. I believe the U320 is in that same family. I am not sure what the difference in the 310 and the 320 is.

The J310 is the plastic version and sell for less than 20 cents. It is an excellent low cost J Fet.

Arrow still has the U310 listed for $10.43 but they have a 200 part minimum. I think I have a couple stashed away somewhere.

Crystalonics has some very good JFETs  - I just bought a CP666 from them this week. Very high dynamics range FET

http://www.crystalonics.com/jfets.html

[WA1GFZ]

A ten tek radio used a number of U310s in parallel. I seem to remember 6 in push pull parallel. CP643 or CP666 come with a good price, yes very high dynamic range. Cubic used a lot of them in their front ends and first IF stages.
Premium RX had a free download of the 8888 manual if I remember. I have a copy somewhere if you can't locate it.

[WA1GFZ]

http://premium-rx.org/reference.htm

complete manual

[KA3EKH]

Need to find something that can replace U320, looking at the data sheet have to wonder what I can use. Most of the time when I need a replacement part just go to EBay and order it but not seeing anything that looks like that, epically the current and power ratings.  That FET is located right before the input of the first mixer on the output of the band pass filter assembly. Don’t know why but there appears to be no damage or issues with the input filter or any of the band pass filters in the radio. Did a check and see no significant loss in the filters but that FET on the input of the first mixer was shorted. Not having the book and no desire to open the filter assembly would assume that all the filters are passive and maybe the blown FET was the first active stage in the receiver. The WJ-8888 has a huge selectable set of band pass filters before the first mixer unlike the WJ-8716/18 stuff where the antenna just pass thru a 0 to 30 MHz filter and hits the first mixer directly, first time I worked on one of those thought someone must have stolen out the filter assembly but looking at the manual shows that’s how there built. Speaking of manuals, the Premium -rx site link for the WJ-8888 stuff don’t work has not for some time. I did send them an email asking if they had an updated address but never got anything back. Maybe I am just not premium enough to warrant a response? All there other links work, just not the ones for the WJ-8888.  Also I feel that I have some additional issues with the input converter beyond the blown FET, disconnecting the 10.7 high IF output of the input converter results in the noise floor of the receiver being very low. When I check noise floor I crank up the gain AF and RF gain all the way up and see what I get then. When I connect the input converter back to the IF input the noise floor rises at least 10 or 12 dB, this has always been a problem with this radio even before I transmitted into it so do not know if there is a issue in the input converter or if these early solid state radios just had lousy noise figures or what? I know from experience that the newer solid state receivers have very low noise floors but also have radios like my old HRO that has a surprisingly good noise floor and sensitivity for something that was produced back in the age of dinosaurs. All this stuff about noise floors and sensitivity seem kind of pointless when operating on 160 AM but I am getting tired of stations telling my they copy me just fine and I can barely hear them.
 

[w3jn]

Paragraph breaks are your friend 

Here's another source for the WJ8888 manual http://bama.edebris.com/manuals/watjohn/wj8888b

You're right, I tried downloading some of the WJ8888 pages at premium-rx and none of them worked.  Have you joined the premium-rx group?  There's a mailing list and you could ask all manner of WJ experts on the list.

[WA1GFZ]

3 J310s or u310 in parallel should work if you don't want to buy a CP643 or CP666
A single one would work with reduced dynamic range. Just check the drain current to make sure the current is ok. I would think 10 ma/ device is a good save operating current.
U320 test current is 27 ma.
I have an electronic copy of the 8888 around the house somewhere if you can't locate a download. I would think the input section is passive followed by a grounded gate amp. Cubic did the same thing with the CP643.

[WBear2GCR]

Paralleled JFets will have lower noise than a single one too... I'd have no concerns about paralleling them to get the required IDSS. It would be a good idea to get units from the same "batch number" they are usually grouped by IDSS batches. Better still would be to static match from a number of candidates for best matches, but probably not required.

An issue might be a metal can vs. plastic pak in terms of shielding.
The data sheet should show where the can is connected.
You can fab an aluminum can or get a derelict other transistor and take it's can, some appropriate goop and shove 'em together... copper foil is a candidate of high potential... of course if the metal can has no benefit in the circuit no worries! The other thing is that a wrap of copper foil (tack soldered) will keep them thermally tracking together too. But that too is a fine point.

I think Siliconix made some U series devices too, and some of them were used in the front end of Tektronix 100MHz + scopes... I think any JFet with similar specs will do just fine - some newer devices simply did not exist when that receiver was made. I'd sub in, need be.

Can we see pix of the innards?

                    _-_-bear

Datasheet: http://www.datasheetarchive.com/U320-2-datasheet.html#

[Pete, WA2CWA]

I didn't compare the specs, but the MPF-102 is a good RF front-end JFET. Mouser has them for .12 each.

[KA3EKH]

The plan so far, ordered a batch of J310 N channel FET on EBay and when they get here stuff one up the front end of the receiver. Looking at the specs for the original U320 and the replacement J310 I am a little concerned about current. The 320 is rated at higher current then the 310 but have no idea if this will be a problem or not.
This is yet another example of my most amazing use of MS Paint to produce something that looks like a schematic. The B+ is 15 Volts and on the other working FET stages in the Input converter, there are two identical sets of amplifiers for the high IF and the output of the second IF the bias on the collector of the 2N2222 is 1.25 Volts. Maybe be best to put two devices in parallel although can picture that doing all sorts of weird phase stuff.
Will do a second post of a couple pictures of the radio next.

[KA3EKH]

Pictures of the radio. The input converter is pulled out and has its cover removed. That section has enough cable to allow it to be opened and placed on its side while in operation you just have to be careful because the DC and AGC inputs are all on the bottom and if not careful you short them on the case.
Each stage of the converter is in its own little sealed box for isolation. The original U320 was in a metal case in a transistor holder but the new J310 things will be in epoxy case but being that the entire input stage is sealed do not think this will be a problem, but if it is it won't be the first time I was wrong about something.

[The Slab Bacon]

Ray,
       Your workshop is entirely too neat and orderly!! How do you get anything done? ? ? ?   

[W4AMV]

The 320 appears to have a slightly higher gm and is probably a larger device. Consider if you can, matching the Idss of your batch of 310s when them come in and simply use two of them in parallel to build a larger FET.

Alan

[WA1GFZ]

I remember when my bench was that clean.....about 20 years ago
The 8888 looks like a well built radio, I wonder how clean the synthesizer is for phase noise. I was surprised when I checked out the 8716.
That transistor clamp may have provided some heat sinking for the device as well as mechanical support. Maybe Siliconix had a numbering system Die size/number of dies in parallel/0 so maybe the guts of a 320 is 2 310s in parallel.

[Rob K2CU]

That RF module looks just like the ones we used to build up out of silver plated double sided PCB stock. Slick.

[KA3EKH]

About five years ago went thru a pallet of the WJ-8716 receivers for someone selling them. They are not bad radios except that they had issues with the tantalum capacitors shorting and the memory batteries dyeing. Weird thing is that out of the twenty or so I worked on about half did not have the filter assembly for the front end of the receiver. They had a band pass filter that connects directly to the first mixer, that’s all. There is an option for a filter assembly but that’s an option. Also they have the weirdest method of selecting IF filters with three of the filters being on the 455 IF and three being on the 10.7 IF. The 8888 is a simpler and older design, don’t know why the number is higher but the design is older, go figure. Attached are pictures of the version of the 8716 with front end filter, without and what the radio looks like. I use to buy 8716 regularly and sell them but the stupid prices they go for today at government auctions put an end to that.
Ray F

[WA1GFZ]

I wonder if that preselector is tunable and could be used stand alone as a front end for an SDR RX?

[KA3EKH]

No, there not that smart. They work in sections of bands and are just switched in and out for the appropriate band. Don’t remember exactly but one covered 1.0 to 3.0, the next one was 3.0 to 7.0 and so on. Think the widest filter was the top band that was 12.0 or 14.0 to 30.0 The 8716 used little reed relays to select filters and think that’s what they use in the 8888 I worked on VHF/UHF surveillance receiver before that had similar set of filters and for selecting filters and preamps they used switching diodes in the front end and they would go bad and generate wide band noise, not on the band your on but usually on a different band. That was a fun project. I always liked the idea of reed relays after that. Biasing diodes for switching signals is too complicated for me.

[WA1GFZ]

Sounds like the Cubic set up. They used pin diodes above 2MHz and relays below. A CP643 grounded gate RF amp. Input had a sense circuit to open the antenna line if the RF coming in was too high level.

[N8ETQ]

Ray,
       Your workshop is entirely too neat and orderly!! How do you get anything done? ? ? ?   

     I've been looking for a few dual gate jobs for a FT-101B my self. Found this
guy but have not dealt with him yet.

http://www.angelfire.com/electronic2/index1/index.html

    My shop is pretty sad right about now, service work V construction along
with the maid's day off adds up to quite a disaster.

/Dan

[KA3EKH]

I installed the new FET. Used a J310 and it appears to be working. Has plenty of gain but I am concerned about the lack of voltage on the source leg, the connection that goes to the collector of the 2N2222 that sets the bias on the device? On all the original U320 devices they have 1.2 volts DC on the source and around 15 volts on the drain, the J310 has 15 volts on the drain but only around 0.2 volts on the source. The 2N2222 is not shorted and has 1.25 fixed bias on its base. Was the bias voltage originally determined by leakage of the old N channel JFET? And because the new JFET has less leakage there is less bias? The stage appears to have gain and low noise so do I need bias on the source leg? To just try this out and see what would happen I just tacked a signal FET in the circuit and after seeing that it did not get hot or have any issues went on from there and I never got around to measuring the current thru the device.
Also still looking for the schematic, tried the second source but resolving the mystery of .djvu files appears beyond my capabilities.
RF

[WBear2GCR]

You have to download a dejavu reader... seach, find, download, install.

The circuit is a cascode. The IDSS of the device will determine the Vdrop created after that resistor that supplies the "input" side. It is probably a good idea to see where the current is - since at high levels I think you want the device smack dab in the center of its class A range so that you don't get clipping in one signal swing direction vs. the other. Otoh, it probably is sitting smack dab at IDSS... unless there is some self bias on the source side, I am not looking at the schematic posted as I write. So you might want to adjust the supply side, the resistor might want to be made higher so that there is more vdrop at the same current draw. I think that is correct.

Or just put two in parallel...

                      _-_-bear

[WA1GFZ]

Your source voltage will come up when you add more j310s in parallel. Don't mess with the bias it may damage this lower current FET. You will just lose a bit of dynamic range with a single j310.

[vincent]

I have used in the past (mid-seventies!) the POWRFET (Teledyne Crystalonics), how it was called at that time, type CP640 (and CP650/51) with excellent results. I remember that these were quite expensive.
A “good” but cheap one is the BF244C.
Scans enclosed.