Author Topic: Replacement for Fluke 700013 IC (quad SPST analog switch)  (Read 480 times)

0 Members and 1 Guest are viewing this topic.

Offline richipedia

  • Newbie
  • Posts: 1
  • Country: de
Replacement for Fluke 700013 IC (quad SPST analog switch)
« on: May 04, 2018, 01:46:45 am »
Hello everybody,

I recently bought a broken Fluke 8842A bench multimeter. As it turns out, two Fluke specific ICs with Fluke stock number "700013" are broken. Those are quad SPST analog switches with some integrated logic (latches and NAND gates) to accomplish the input voltage scaling and the current source switching for the resistance measurement - see images below or see the users manual at assets.fluke.com/manuals/8842a___imeng0300.pdf.

I tried to find a genuine replacement IC, but was not successful. From the 8842A users manual, I can deduce that Fluke stock number "803478" is a similar device, but I have no success either in finding one of these. After so much frustration, buying another broken Fluke 8842A or Fluke 8840A (both contain said ICs) seems too few of a challenge to me  ;), so why not design a replacement, that can replace any of the five 700013s in use in a 8842A? And why not share it with you? I already figured there are some guys also looking for a suitable replacement.

The logic functions can easily be replicated using standard 74xxnnn CMOS logic gates, I believe. As far as I'm aware, I need to consider for those...
  • supply voltage
  • propagation delay
  • digital logic level voltages (inputs need to match levels at hand on the PCB, outputs of the (N)AND gates need to match levels required by the analog switches)
  • latch must be transparent while the signal not(L) at pin 3 is low

The analog switch function can be implemented using analog switch ICs (who would have thought?). Jelly bean ICs like the Intersil DG411 or the Analog Devices ADG411 (better specs) are readily available for hobbyists like me, but I'm not familiar with all the peculiarites of these devices, I believe. Anything to consider for this purpose other than the following?
  • supply voltage: must run off analog voltages -8,2V / +7,5V and -5V / +7,5V and 0V / +15V
  • current consumption
  • propagation delay
  • on-resistance
  • off-resistance a.k.a. leakage current
  • charge injection

The values of most of the requirements above need to be extracted from the PCB and from the two working 700013s I have. This is where I'm at so far. Just start screaming at me if there are some horrible mistakes :)
  • digital logic levels: using my humble DS1054Z, I measured clean square waves at the digital signals with low levels close to 0,0 V and high levels at 4,0V or above.
  • propagation delay: tbd
  • latch transparent while not(L) is low: should be easily obtainable using a standard 74xx373 or 74xx573 and an inverter in front of the latch input
  • current consumption: if I draw more than 4 mA from either the -8,2V rail or the +7,5V rail, the voltage drops dramatically. Haven't tested the +15V rail yet. The +5V rail seems to be very stable - no wonder as it is supplied by a standard 7805 voltage regulator IC.
  • on resistance: plots in the DG411 data sheet show that on-resistance is dependent on the bias voltage at the switch nodes. So I closed a 700013 switch, put a DC voltage on one of the switch pins using a trim pot, and measured DC resistance using a measly multimeter. Is that sufficient? Seems like a very casual, unsophisticated approach to me.
  • off-resistance a.k.a. leakage current: similar to the above, I opened a 700013 switch and measured DC resistance using my measly multimeter. Result: infinite ohms. Bummer, leakage current is too low. Next try: put a high voltage on one of the switch pins, connect the other pin to the multimeter in DC-voltage-mode connected to ground. From the measured voltage and the (not very precisely specified) input impedance of the multimeter, I can calculate that leakage current must be in the order of 10 pA
  • charge injection: there is a nice Analog Devices appnote showing how to measure charge injection. Put a cap with known capacitance at one of the switch nodes, open and close the analog switch using a function gen at a rate of 10 kHz, and track the capacitor voltage using an oscilloscope. There should be a significant voltage rise when the switch is turned off. Result using a 10 nF cap: there is no voltage rise. Apparently, the capacitance is too large. Using no capacitor at all (so that just the probe input capacitance affects the circuit), I can see and measure the charge injection voltage rise. Any ideas on how to measure probe input impedance?

Well... enough ideas and questions for now. Or is it? Ah, wait!

Bonus question: what might be the root cause for the 700013s failing in so many Fluke DMMs? Is there a way to prevent my little replacement from suffering the same painful death? I would guess that there might be an unfortunate condition at power up, causing multiple switches which act on the same node to be closed at the same time, resulting in a short circuit.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf