On Sat, Sep 17, 2022 at 11:50 AM, Steve Ratzlaff wrote:

The Q1 input transistor makes all the difference. I've been trying various Q1's and have found several that work very well. Low end input impedance is now under one ohm. The same 2N5109 Q2 is used; I haven't tried changing it (Central Semi, DC beta 110). I've been looking at the 1 MHz point for my Q1 comparisons. Best is 1.05 ohms with an obsolete Motorola 2N4401; a metal can BC108B gives 1.07 ohms; a metal can 2N3947 gives 1.10 ohms; an obsolete Motorola PN2222A gives 1.11 ohms.

?? I note the MPSA18 has a HFE of 500 to 1500, will this work as the first transistor to lower output impedance even more?? FT is 100MHz.
????????????????????????????????????????????????????????????? Mikek

Before I put time in making a documentation package (will take me couple of
days) I would check if it is interesting for somebody -- don't want to waste
my time on something nobody needs.

It is Keithley 2001 upgrade. Older instruments had a single PLCC-44
HN27C4096 EPROM for their firmware. That firmware ended at A08 version. Then
Keithley made a newer instrument with firmware in 2 8-bit EPROMs and
released Bxx version firmware with greatly extended functionality. The
latest (and almost certainly the last) version is B17. Bxx series firmware
is too big to fit in that HN27C4096 and that was the reason they made a
newer 2001 version. Those newer instruments have everything identical to the
older ones except a new Digital (CPU) board. That board, in turn, is almost
identical to the older one except having 2 4096 Kbit 8-bit EPROMs (for 8
Mbit total) with some circuitry to interface them to the same CPU instead of
one 16-bit 4096 Kbit EPROM in the older instruments.

I made a small 2-board stack adapter for a single 16-bit 8192 Kbit 5V FLASH,
MX29F800CBTI, that replaces that single 4 Mbit EPROM. It is installed into
existing footprint of that PLCC-44 socket (which needs to be removed,
desoldered). MX29F800CBTI is 2x the older EPROM size, exactly as that 2-chip
solution on the newer boards so it fits the B17 firmware just fine. As it is
16-bit FLASH no additional circuitry is required, just ODD/EVEN parts of
firmware have to be assembled into a single image.

I actually made it 2 years ago but didn't have time to install it until
yesterday. B17 firmware runs just fine, everything fits mechanically without
problems. I built one panel of 9 tops with thin TSOP48 sockets for that new
FLASH 2 years ago, now one of those tops (as it's been said it is a STACK of
top and bottom boards) along with the bottom is in my 2001.

The MC29F800CBTI is readily available from DigiKey and it is not expensive
($5 or so for single quantity). I do still have 8 assembled tops left that I
don't need along with matching bottoms. Also have 9 FLASH chips that I can
program with B17 firmware if there is interest.

I actually have even more of those boards (4 more panels of 9 each) but
those are bare boards, not assembled, if somebody needs them. However, those
should have FLASH chips soldered to those tops directly, without sockets.
The reason is those sockets are rare as hens' teeth and I don't have more of
them except those remaining 8 that are already assembled. As a matter of
fact I DO still have sockets but it is not the SOCKETS that are rare, it is
their COVERS that lock the chip in the socket are. I don't know why but they
decided to supply socket bases and their covers SEPARATELY, as separate
parts and those covers are unobtanium now.

I don't think it is a big problem to have the FLASH soldered in as there
will be no more new firmware available but anyway...

It is relatively easy to upgrade an older 2001 if one have proper
desoldering equipment to remove the existing PLC-44 socket without damaging
the board. However, there is one difficult step is one doesn't have
precision soldering iron and microscope to work with. It is that yellow wire
clearly seen on the attached photo. It is A19 signal from the CPU required
for addressing the bigger FLASH. This signal is not routed to anything on
the older boards, CPU pin is just soldered to an unconnected pad. Not
actually something impossible, something like 10 minutes job from start to
finish at most for somebody who is used to such kind of jobs but might be
impossible without proper equipment and steady hand.

B17 firmware runs OK on a modified older instrument (mine has rev.H digital
board and 0548310 serial number), everything works, re-calibrated like a
charm without any problems today using my Wavetek/Datron 4808 and Keysight
3522B FG for the last step in Low-Level Calibration (2Vrms @ 1Hz -- 4808
can't go that low in frequency). The only quirk was it requires +2.0ADC in
step 14 and 4808 without external amplifier can't source 2A. However, it can
source 1.9999999ADC that should be close enough :)

I will upload all design files and some pictures from the upgrade process if
there is interest. No reason to bother if nobody is interested.

Will also sell those boards including those remaining assembled tops with IC
sockets and FLASH chips pre-programmed with B17 firmware.

BTW, it tells it is 2001M, not just 2001 with that B17 firmware :)


P.S. Sorry for attaching the photo instead of putting it into files -- don't
want to waste the file space if nobody is interested.

---
*
* KSI@home KOI8 Net < > The impossible we do immediately. *
* Las Vegas NV, USA < > Miracles require 24-hour notice. *
*

Added to:
New T50-2 36T measurement is a V-Meter Bandwidth Q measurement and a
SimSmith model demonstrating Injection Transformer fixture loss when
driven with 50ohm generator. Note V-Meter voltage ratio measured Q=205,
BW Q=207. SimSmith model suggests DUT with Q=207 will result with
voltage ratio Q=204.

John KN5L

Jacques, Thanks, makes sense with full scale actually being 3.0. I'm used to AC voltmeters full scale actually being 3.16 even if only marked up to 3.0.

John KK6IL

Evaluating frequency range can be performed by removing DUT from
Injection Transformer fixture, adding a wire between input and output of
Injection Transformer fixture and evaluate output of RF voltmeter.
FT50-43 50 turn, fixture with prototype RF voltmeter as shown, 1dB
bandwidth is 7kHz to 14MHz.

John KN5L

Success! Some emails ago Jacques mentioned using a capacitor divider at
RF voltmeter input. It may be a requirement!



is updated with new RF voltmeter schematic and Q measurement results. RF
voltmeter input coupling capacitor is 10pF with a 120pF shunt added to
FET Gate.

RF Voltmeter calibrated using DG1032. Rather flat from below 100kHz to
10MHz, with 1dB down at 10MHz.

Demonstrating direct reading Q meter with Q=250 at full scale. Measured
DUT is well within range for the device.

John KN5L

John,

The attenuator must adjust levels according to the Q range selected.
To go from a Q range of 30 to 100 requires a level change of 100/30 = 3.333
Attenuation in dB = 20 * log(3.333) = 10.46 dB

To go from a Q range of 100 to 300 requires a level change of 300/100 = 3.000
Attenuation in dB = 20 * log(3.000) = 9.54 dB

Note that to go from a Q range of 30 to 300 requires a level change of 300/30 = 10
Attenuation in dB = 20 * log(10) = 20 dB = 10.46 + 9.54

Jacques,? VE2AZX
ve2azx.net

Nowadays, fixed attenuators are available for pennys in SMT packages. Don't know what was available in 1975, and the photos of the A3 assembly in the on-line manuals aren't clear. R2 and R4 have the same part number so wouldn't work to have an extra 50 ohm series R included.

I'm used to signal generators having fixed output Z's the match the loads (and cables). I guess the attenuators will provide the proper attenuation into a 50 ohm load regardless of the driving Z, but too lazy to prove it.

I don't understand the 10.4 and 9.6 attenuator values. 0.4 db error may mean much measuring audio signals, but a 5% error measuring Q is not trivial. More of HP's black magic at work here, I suppose.

I like the concept of an attenuator providing 10 db steps for meter ranging, and I have a 75 ohm step attenuator that may finally find a use.

John KK6IL

开云体育

I should have noted a small nit to pick with swallop's answer: Q1 is not a variable load. It supplies the bias current for Q2. More current = higher amplitude (up to a point). That's the relationship exploited by the ALC loop.

--Tom

-- 
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070

On Mon, Sep 19, 2022 at 12:16 PM, <swallowp@...> wrote:

A1 Q2 operating as a grounded base oscillator.? Q1 is a variable emitter load which regulates the amplitude of oscillation, driven by the ALC control voltage from Q12 on A8.

?

?Thanks, I see it now.
??????????????? Mikek

?

开云体育

Renee got it right. Q1 is a common-base (non-inverting) amp, around which there is a positive feedback loop formed with a tapped inductor and capacitor. Q2 is a current source for biasing Q1. This current controls the oscillation amplitude (more current = more amplitude; the relationship is pretty linear over a wide range). The current, in turn, is set by the ALC loop to keep it at the set level.

Just for fun, now that I look at the oscillator, where are the active components, I see the tuning, but what makes it oscillate?
??????????????????????????????????? Thanks, Mike

Updated HP4342A schematic, includes ALC and Oscillator.

John, 50 ohm series input may be built into R2 attenuator input series
resistor. Don't know, as attenuator resistor values are not marked.

In "30 3" switch position, connection is straight through, impedance
matching is not required, as long as load is near 5o ohm, using A4 Assy
R1 parallel with Q1 Base.

John

Yes, exactly why I did it, the schematic I'm using doesn't have them in order, Pages away from each other in fact, so I had to keep
scrolling to find them, so I? put them together. For those unfamiliar with Dropbox, the schematic opens at 28% of size, (at least for me)
the box at the bottom allows you to enlarge and the slider bars move it side to side and up/down.? I do like that djvu schematic quality.

>>Q3 & Q4 of the power amp are identical circuitry to the impedance already analyzed to have a very low output Z,
so I would expect to see a 50 Ω resistor between the emitter of Q4 and J1 to make the power amp output match the input Z of the attenuator.<<

? It would seem it should have a series 50Ω, but it's not there and, to make the attenuators work right, assuming 50Ω pads, it should be.
For added info, R5 on A1A2 'RF Power Amplifier Assy' is a 51.1Ω resistor, (it's not clear on the schematic I posted).
? I don't know the input impedance at the base of Q3 ('A1A2 'RF Power Amplifier Assy'), I do see it is higher than on the 'Impedance Converter" injection transformer driver
transistor Q1, because of R10, 147Ω in series with the base of Q3. I doubt it would increase the output impedance enough to make the A1A2 'RF Power Amplifier Assy' have a 50Ω output. that analysis is beyond my ability, If it's not 50Ω, it's perplexing, So, I post to learn from others.
???????????????????????????????????????????????? Thanks, Mikek

We have 1 GB of storage space at?--

Thanks for the combined schematic - hard to follow flipping through pages of a PDf.

The input impedance of the Impedance converter is 50 Ω (R1, R2, R3 in parallel). The steps of the attenuator would be sized to present 50 ohms at it;s input regardless of the range switch settings.

Q3 & Q4 of the power amp are identical circuitry to the impedance already analyzed to have a very lot output Z, so I would expect to see a 50 Ω resistor between the emitter of Q4 and J1 to make the power amp output match the input Z of the attenuator.

John KK6IL