This three-wire pressure transducer pinout, drive type, and electrical specifications are unknown. It should be a typical industrial transducer. I have tried assuming it is an amplified device with a 0-5, 0-10 V output, or 4-20 mA output. No combination of wiring results in an output signal correlating to applied pressure. Edit Note: I applied 10 bar, which the Agilent 1100 HPLC system easily resolves in normal operation.

Pulse dampeners

The transducer comes out of a late 1990's designed Agilent 1100 HPLC scientific pump, where it measures the output pressure up to 440 bar (6500 PSI). It is part of a pulse dampener, so I wish to keep it intact. I do not have access to the drive PCB any longer.

The markings are: Agilent Technologies Part No. 5065-9907 Revision A Date code 21/03 PA-23R / 81088 440 bar

I suspect this may be a Keller OEM transducer. Keller's currently advertised lines do not include a 23R, but they follow a similar product naming convention - PA-23R could make sense.

The transducer has three wires in a 0.100 pitch ribbon cable, so unshielded, One is black striped. It is pin 1 in the tables below.

I've monitored the drive (edited) current while powering the transducer for repetitive pulses as evidence of a microcontroller. I also monitored for pulses on the output and power fluctuations during boot time. No evidence of active electronics.

Pulse dampener with pressure feedback

The only configuration that shows any response at all is #5 below. The output voltage decreases slightly as pressure increases. However, the output voltage depends on the supply voltage. This seems unusual if it is an amplified transducer.

Note: During each resistance check, I cycled the applied pressure between 10 bar and 1 bar (atmospheric).

The pin-to-pin resistance matrix is:

pin 1 (black) pin 2 pin 3
pin 1 x 1.52 MΩ 13.98 MΩ
pin 2 1.52 MΩ x 11.12 MΩ
pin 3 13.94 MΩ 11.08 MΩ x

The pin-to-pin voltage drop is:

pin 1 pin 2 pin 3
pin 1 x 1.64* V open
pin 2 0.73 V x open
pin 3 1.93 V* 1.72 V x

*Values decreased when attaching the meter.

Powering the transducer with a 15 VDC supply current limited to 60 mA gives this result set:

Config pin 1 (black) pin 2 pin 3 current (mA)
1 +15 0 out 43 mA
2 +15 out 0 16 mA
3 0 +15 out 22 mA
4 0 out +15 0 mA
5 out +15 0 0 mA
6 out 0 +15 0 mA
  • \$\begingroup\$ I googled Keller and found the following datasheet, manual, and spec: (1) download.keller-druck.com/api/download/feHh39CHy7AhquSGgv3UCS/…, (2) download.keller-druck.com/api/download/8p9KraBomWGhZsPhncEEc/en/…. (3) Electrical Connection - GSP Plug EN 175 301-803-A (DIN 43650) \$\endgroup\$
    – tlfong01
    Commented Feb 9, 2022 at 2:23
  • \$\begingroup\$ Throw it in the garbage and buy a new one is my strong recommendation. \$\endgroup\$
    – Andy aka
    Commented Feb 9, 2022 at 9:09
  • \$\begingroup\$ @Andyaka, I have many of these, and they're for test rigs to be shipped to remote workers. Also, there's a lead time problem for this type of transducer. \$\endgroup\$
    – ndemarco
    Commented Feb 9, 2022 at 17:43
  • \$\begingroup\$ which pin is the black wire? it looked almost like a constant current device... might be worth taking a couple of extra measurements: capacitance and resistance to the outer metal body. and possibly an extra pic closeup of the back where the wires enter the blue potting. \$\endgroup\$
    – Abel
    Commented Feb 10, 2022 at 0:35
  • \$\begingroup\$ Pin 1 is the black wire. Good call - edited the question. \$\endgroup\$
    – ndemarco
    Commented Feb 11, 2022 at 4:03

3 Answers 3



Keller 23 Connection

Keller 23


(1) Keller 23 Datasheet

(2) Keller 23 Connection

  • 1
    \$\begingroup\$ this doesn't match the current measurements. \$\endgroup\$ Commented Feb 9, 2022 at 3:57
  • 1
    \$\begingroup\$ @tlfong01, this is what I said I tried within my question (0-10V and 4-20 mA). \$\endgroup\$
    – ndemarco
    Commented Feb 9, 2022 at 17:45
  1. If you carried out your measurements with the sensor under a few different pressure levels (say 0, 200, 400 bar) showing all the results could provide insight.

  2. The symmetry of table 1 suggests a passive sensor (even though the high ohms values of the very same table suggests the opposite: active electronics!).

  • If sensor is a variable resistor (maybe 3rd wire is a ground just for noise reduction), or, a dry potentiometer, these cases should be revealed by the measurements I suggested above.
  • Ditto if sensor is capacitive (again only 2 relevant wires). Apply a sinusoidal AC to a pin and connect another pin to a resistive load and measure VAC.
  • Try to learn if or which passive sensors were in use in the early 90's, especially by Agilent & Keller.
  1. I didn't understand your

I've monitored the drive voltage while powering the transducer for repetitive pulses as evidence of a microcontroller.

  • If sensor has active electronics with digital interface (open collector output, I2C, UART (USART) / RS-232, RS-485), ofc it would be tough to reverse engineer the RX/TX protocol but if you connected two pins to a DC supply (like you did before) and scoped the 3rd pin (first without, then with, say a 10k pull-up resistor between it and your Vcc, then a pull-down), and also making & breaking the Vcc connection a bunch of times, you might observe some digital activity.
  • Again it would be useful to know which 3 wire interfaces were in use in the early 90's, especially by Agilent & Keller.
  1. It's not super likely 15V would fry such a unit, but possible (unit could be a custom design for a very specific application which entailed it's being quite fragile), so it might be worth grabbing a second unit and keeping Vcc to 5V.
  • \$\begingroup\$ I'll do a better job of applying varying pressures. I applied low (10 bar) pressures in the readings above - no change was noted. I've updated the table to include that data. I sneaked up on the 15V, starting at 5V. After applying 15V, I re-checked the 5V numbers, and they were unchanged. I also checked multiple sensors at 5V - all behave quite similarly. I have some work to do tomorrow. Thanks for the suggestions. \$\endgroup\$
    – ndemarco
    Commented Feb 11, 2022 at 4:02

Before you do anything else, go to a local dentist office and pay them for an X-ray of the sensor, in three radial orientations, roughly 120° apart - mark it on the sensor. That should give you a good idea of what circuitry is present inside. They may need to use a 2-5x longer exposure, depending on how thick the steel is. Anyway, digital X-ray sensors are popular, so on those the results are immediate and exposure can be adjusted quickly.

In the configurations that consume power, you’re assuming that the output is voltage. Look at it with oscilloscope: what is it? Does it oscillate? Is it stuck high or low? The output could be current, with a separate supply input. In such case, you’d want to attach a resistance from output to either + or -, and see what voltage you get.

You also should have some idea of the impedance between the pins - complex impedance, at least at a couple of frequencies, and with a couple of DC bias voltages applied. In other words, using a digital RLC meter with DC bias option. The resistances at very low voltages are informative but don’t provide much insight. It’s probably not a passive capacitive sensor, since those should have much more than MOhms between terminals. It’s also not a passive inductive sensor.

A low current curve tracer like the Mr Carlson’s Lab one would be very helpful to see what’s connected to those pins.

For identifying the common/0V pin, it’s worth checking if any pin is connected to the case, perhaps though a “solid” capacitance that would show up when measured. It’s easy to tell apart a real 1nF capacitor from a semiconductor input that happens to provide some capacitance multiplication action and mimics one - as long as you can measure complex impedance and/or can do curve tracing.

There’s nothing unusual about ratiometric outputs, and in fact those would be preferred to a calibration related to a fixed reference voltage. But I doubt very much that the “response” you obtained in config #5 is the real deal. Likely it’s a parasitic path though active electronics. I sometimes see similar behavior with resistive sensors connected to unpowered instrumentation amplifiers.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.