On Nov 18, 2020, at 1:53 PM, Ed Breya via groups.io <edbreya=yahoo.com@groups.io> wrote:

For some reason I've been referring to the original working JFET as a J310, but looking back all the way to the OP, I'm not sure what it is - all kind of parts are floating around this thread. So Charles, can you identify the one official, working JFET that I've been referring to?

The reason I ask is that I looked up the J310, and found it is a very heavy beast - VHF/UHF, with Idss 24-60 mA - way too big and fast for this little scope amplifier, and way too big for Tom's estimated max of 30 mA for circuit operation. The correct part must be something else more "normal," among many possible choices. The J309 may work, but even it seems a bit much.The J300 seems about right, and the J210 or preferably J211 may be a good option. Also, I mentioned the 2N5485 and 2N5486 earlier, which I'm pretty sure were used a lot under various Tek part numbers. I think especially the 2N5486 was used extensively in scope input amplifiers.

Anyway, there are types that it could be, and types it shouldn't be.

Ed

On Nov 18, 2020, at 17:17, Ed Breya via groups.io <edbreya=yahoo.com@groups.io> wrote:

The cleaned up 7A42 is still working OK. I'm calling this one fixed and done, ready to go back on the shelf. Hopefully it will still be good whenever I need it.

Ed

Ed,

I’ve analyzed the circuit. Any IDSS below about 30mA will satisfy the
loop, as I’ve noted twice before. If much lower than that, dynamics may
suffer, but the upshot is that there’s a very wide window.

It’s that accommodating nature that leads me to suspect that the OP may
have other problems.

Tom

Sent from my iThing, so please forgive brevity and typos

On Nov 18, 2020, at 12:15, Ed Breya via groups.io

<edbreya=yahoo.com@groups.io> wrote:

There may be some confusion over "selecting" versus "matching," due

to the different input amp topologies also included in the discussion.
In this case, it's an opamp-stabilized system, with only one JFET up
front, for each channel. There is no need for matched JFETs, but
apparently there is a need for the JFETs used to fall in a narrower
Idss range than the generic part types they come from. It looks like
the official Tek part number used is provided by the vendor from J310s
or similar, selected for some range of Idss according to Tek's spec for
this part. The problem is that we don't know the selection criteria
applied, but we do know the one good original J310 found in the circuit
appears to work, so finding another VHF JFET that has similar Idss
should work too. We don't know if the original is an ideal example - it
may be at an edge of the desired range, or right in the middle, but at
least getting close to whatever it is should be in the ballpark.

Ed

On Nov 18, 2020, at 12:15, Ed Breya via groups.io <edbreya=yahoo.com@groups.io> wrote:

There may be some confusion over "selecting" versus "matching," due to the different input amp topologies also included in the discussion. In this case, it's an opamp-stabilized system, with only one JFET up front, for each channel. There is no need for matched JFETs, but apparently there is a need for the JFETs used to fall in a narrower Idss range than the generic part types they come from. It looks like the official Tek part number used is provided by the vendor from J310s or similar, selected for some range of Idss according to Tek's spec for this part. The problem is that we don't know the selection criteria applied, but we do know the one good original J310 found in the circuit appears to work, so finding another VHF JFET that has similar Idss should work too. We don't know if the original is an ideal example - it may be at an edge of the desired range, or right in the middle, but at least getting close to whatever it is should be in the ballpark.

Ed

Hi Tom, my thinking went thus::

I understood from previous posts that there were a pair of identical fets. I similarly understood they were "selected".

If these two statements are correct I stand by my original suggestion.
I should admit I have not worked on a 2215 and base my opinion on reading the thread, not looking at the actual circuit.

Regards dave
⁣Sent from BlueMail ​

On 18 Nov 2020, 10:17, at 10:17, Tom Lee <tomlee@ee.stanford.edu> wrote:

Hi Dave,

Keep in mind that this is a 60MHz scope, and that the circuit in
question is a simple buffer. I don't see how adding a new constraint of

matching helps to fix it. I still think that the OP either has a
collection of JFETs that Murphy has arranged to have extremely off-spec

IDSS (hence my advice to simply measure it; it's trivially easy to do),

or hasn't gotten the pinouts quite right (there is not a single
standard
pinout). It is also possible that there is a separate hardware problem
that is making one channel excessively finicky. It's straightforward to

eliminate or verify possibilities 1 and 2, so if it's neither 1 nor 2,
the OP should move on to 3. It's not hard to debug the bias loop, so if

he eliminates 1 and 2, getting through 3 should be fairly quick.

Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 11/18/2020 01:55, dave G8SFU via groups.io wrote:

Can I add a further thought to Ed Breya's suggestions.
By all means use Ed's procedure to get the best possible spec for the

required fet.

But then try and match a pair of new ones. It seems to me much more

likely to work than matching a 40 year old one with a 'new' one.
Regards. Dave.

⁣Sent from BlueMail ​

On 17 Nov 2020, 20:58, at 20:58, "Ed Breya via groups.io"

<edbreya=yahoo.com@groups.io> wrote:

It may be good to rethink and summarize what's going on with these
parts. Here's my take on it, presuming I haven't missed something.

1. There is one "good" original J310 that works properly in either
channel. This indicates the circuits for both are likely OK, as long

as

the "right" JFET characteristics are present.

2. The main issue is finding another JFET - a J310 or similar one -
that has the "same" characteristics as the good one. Alternative

part

trials so far have been unsuccessful.

3. The input circuit of the 2215A is of the opamp-stabilized type,

so

should be quite tolerant of JFET DC characteristics.

4. The "proper" part is 151-1124-00 = Siliconix J2400, which

apparently

is not a standard commercial part.

5. You have to get the pinout right, and we're assuming a TO-92
package.

6. Measuring Idss

Recommendations:

1&3. You can be pretty confident that both channel circuits are OK.
However, it's possible that one may have a minor fault or part out

of

tolerance such that it's OK with the good J310, but another part

that's

close and should work, doesn't quite. For now I think it's safe to

say

both are OK, and can be determined once a "right" part is found.

2. All along, the discussions have been mostly about getting Idss

right

- this is still the case, for DC conditions, one way or another. If

you

can find a VHF-class JFET (more on this in #5) that you can select

to

be close to the good J310, then it should work.

4. Let me explain a little about Tek and selected parts. In the old
days, when particular characteristics were needed, if they could be
selected out from stocks of generic parts at reasonable yield, Tek
often did this in-house, and a dash-number was assigned to the base
part number. This assumed also that the generic parts were used in
sufficient quantity that those not selected out (usually the -00)

would

be used up elsewhere. Later, as the real cost of all the in-house
selection and inventory and tracking complications became evident,
there was a big push to reduce this sort of stuff. One way is to

have

the vendor pre-select for the desired characteristics, out of their
much larger volumes, and assign a custom part number. This can be

done

if you're a large enough customer, and you pay a little more for the
parts to cover the vendor's cost of doing it. The benefit is that

you

get what you need, under a single part number, and you know the true
arm's-length cost. So, the supposition that J2400 is a custom part
number is likely correct - it's selected by the vendor from one of
their generic types, and sold only to Tek.

5. The pinout is important - more than just getting it connected

right.

What you want is a TO-92, with the source lead in the middle. VHF

parts

typically have this arrangement to minimize proximity and

capacitance

between gate and drain leads. The exception is if it's for

common-gate

topology, where you'd ideally want the gate in the middle, and it's
RF-grounded, and shields the drain from source. So, in choosing
possible candidates for substitution, first they should be listed as
"VHF amplifier" in the application highlights. Then look at the

pinout,

and pick only those with the source on the middle pin. The drain and
gate on the outside don't matter - you can always flip them around,

but

of course be sure what's what when installing.

6. For Idss, the easiest is to use a curve tracer and measure the
original good one - it's the gold standard, so take good care of it.

If

you don't have a curve tracer, you can rig up a simple bias circuit

for

checking and comparing. The best would be to measure the DC bias
conditions in the actual working circuit. The gate is assumed at

zero,

so all you need are the source and drain voltages, and you can

estimate

the drain current. Then set up a resistor bias scheme that gives the
same conditions if the same JFET were present. Note that this is not

an

Idss measurement (unless the source happens to also be at zero) -

it's

even better - an actual in-circuit test that can be done on the

bench

to sort the best parts to try in the real thing. In reality, the

vendor

likely sorted for a certain Idss range at a certain drain voltage,
which you don't know, but an in-circuit equivalent should be as good

or

better. There is a simple proxy for Idss, that you can use to

pre-sort

parts, rejecting those that are very unlikely to land close. Just
measure the "on" resistance of the drain-source (with gate tied to
either), with an ohmmeter, compared to that of the gold standard. I
think you'll find that the switching type JFETs will be quite low
(10-50 ohms), and the small, slow ones (like for high-Z DC

amplifiers)

quite high (over about 200 ohms). The VHF and HF ones are typically

in

the mid-range around 50-200 ohms.

The idea of putting in a temporary socket is good too, but a little
awkward working down in the guts. That will certainly tell what

works

and what doesn't.

That's all for now. Good luck.

Ed