LMH6321 CL pin wrong & output instability

Hello Gerrit,

Sorry to hear about all the trouble you have had while trying to get the issue resolved. Looking at the circuit, I am a little confused with the bottom-most amplifier which looks like it is trying to set a mid-bias shift. I dont see how the loop for that particular amplifier is getting closed. Is something missing in the schematic?

-Samir

Hi Samir,

That bottom-most amp is just a comparator, so its loop isn't closed.

Gerrit

Hello Gerrit,

Is it possible that you could de-solder the 100kohm resistor connected inbetween the node of 38.3k & 27nF and check to see if the CL pin is returning to virtual ground? This configuration would essentially give us an idea as to whether the previous circuit is working or not in your new pcb. You have the amplifier configured with Rf = 1kohm instead of Rf = 2kohm but it would atleast give us an idea as to which signal path is causing issue. Also, what is the input impedance of the comparator positive pin?

Best Regards,

Rohit 

Do you think a 100k resistor to a DC open circuit (practically) could affect the output?  This is a high-current emitter-follower driver, and it didn't seem to me that that 100k load could have anything to do with it at all.  I will try disconnecting that later and report the results.

The comparator is an MCP6001 from Microchip with input current in the pA range.  My CL pin is +15 V even with no output from the LMH6321, so I don't think it can be due to something in the comparator.

Yes, the feedback resistors between the two versions are different as you say, 1k vs 2k.  That would make a slight difference in the load on the LMH6321, but the working version was fine with two 100 ohm loads on the "output" lines shown in the schem, so it's hard to imagine the feedback resistors would be the cause of the problem.

Thanks,

Gerrit

Hi Gerrit,

Thanks for the comparator info. Yes. I don't think the 1k vs 2k ohm value should make any difference. That's why I wanted to check whether by removing the 100k resistor, the added comparator removed makes a difference in the output loading of the LMH6321 and thus returns the state as previously you had in the previous pcb.

One thing I found out in your drawn schematic is that you have connected pin 5 & pin 7 via a 30.1kohm resistor. I assume that you are using a 7-pin DDPAK package. Is there a reason instead of connecting pin 5 & 6 together, you have connected pin 5 & 7 as shown in the Application circuit? You should be connected to V+ and EF instead of V+ and VCL, right?

Thanks,

Gerrit

VCL in the app note figure is +15 in my circuit, and R3 in the app note is my 30.1k resistor.  CL connected to +15 through 30.1k just sets my current limit.

Pin 6 on the DDPAK package is the EF output, and I'm not using that so I've left it open.

Gerrit

Hi Gerrit,

I understand. I was suggesting to measure the ohm value of the 30.1k resistor, but it seems to me that you have already done that. The main issue seems to be the CL pin that goes to +15V, which I don't see why it should go to that high a value. Could you also please reduce the supply voltage and see if the CL pin is brought down to virtual ground voltage?

Best Regards,

Rohit

Hello Gerrit,

I am wondering if there is possibly a transient over voltage condition on the Vcl pin (or maybe a latch up?).  It is listed as abs max at +-1.2V for that pin.  It is possible that during turn on the input bias current is not sufficient to keep the voltage on this pin from spiking.  Can you find a way to put a capacitor [to ground] on the Vcl side of the 30k resistor?  Something small like 0.01uF should be sufficient, as this would delay the voltage ramp on pin 7 during turn on. 

Also, your schematic does not show any supply bypass capacitors.  Can you describe the supply bypass capacitor configuration? 

Loren

The resistor from +15 to CL hasn't changed between the two versions of the circuit.  It worked fine before, and according to the data sheet it should.  I will try reducing the supply voltage and report.

Gerrit

Hi Loren,

Hmm, that's interesting.  The transient would have to exist in the new AC coupled output version and not the old.  The CL pin is an opamp input, though, internal to the LM6321, and wouldn't putting a cap on it result in instability?

The supply bypass hasn't changed between the versions.  Both the + and - 15 supplies have very low ESR polymer electrolytics right next to the LMH6321 (~1 cm from the power pins) and there are 100n 0603 ceramics right on the power pins.  Both supplies look fine on the scope, too.

Gerrit

Here are some bench notes:

• Varying +/- 15 does not cause CL to go to virtual ground.  It just follows the + supply level, even down below 10 V.
• The 100k resistor to the comparator input is difficult to remove, and I really don't think that could be the problem.
• The 30.1k current limit resistor does read 30.1k with an ohmmeter.
• I'm attaching scope shots of the CL pin on a rev B board (working) and rev C (not working) when the 15 V supplies are first turned on.  On a good board the pin follows +15 until just over 1 V when the LMH6321 internal loop starts working and pulls it down to virtual ground.  On a bad board it begins to work but then fails and follows the 15 V supply up.  The CL pin does seem to be a diode drop or so below the 15 V supply.

Any ideas?  Thanks for the help.

Gerrit

Hello Garrit,

It is strange that with the increase/decrease of supply voltage, the CL pin is not returning to virtual ground. So, you are seeing a diode drop or so below the 15V supply which confirms us that the internal loop of the V/I converter is not working. I am not sure whether you have noticed the supply current of the board, but it would be a good idea to look into that as well, just to know if the part is placed into Thermal shutdown, you would measure the non working pc board current 8mA lower than the working pc board. Also, it might be worthwhile to note the voltage at the EF pin as well, as this might give us some clue as to whether if it is a low voltage, the part might be in thermal shutdown.

Also, I am not sure whether you are doing power up sequencing or not on your pcb, but sometimes we have noticed that, not having power up sequencing on the board upon startup throws off the amplifier, not just LMH6321, into an unknown state. There might have been a latency between turning the supply of internal amplifier in V/I converter of LMH6321 and the input of CL for first time, which might even cause damage to the input of the internal amplifier. This latency might have been small for the working pcb.

It might be worthwhile to remove the 30.1k ohm resistor by keeping CL open and just power on the part as well to check and see if CL tracks the V+ supply. 

Best Regards,

Rohit

Thanks Loren -- you found it.  I thought I checked all the pins, but my ground pin is actually not connected.  In this version of board layout I used an Altium "net tie" instead of a jumper to tie two ground planes together, and that resulted in all my grounds around the LMH6321 not connecting to their local ground plane.

Now I have to figure out if I can rework around this problem, but this is definitely the cause of my CL pin error.  Thanks to everyone for their help.

Gerrit

Hi Gerrit,

I'm glad you found the problem.  I hope you find a good fix.

Loren