LM393: Schematic review

Hi Shu-Cheng,

what is the purpose of C703 and L707? These components can generate a ringing. Is that intended?

Kai

Hi Sirs,

C703 is used for impedance matching.

L707 is used as a high impedance to make the signal not be affected.

Please see as below in details, thanks a lot.

Hi Shu-Cheng,

what signal is arriving at Uniview_RX? Is it a square wave of 6.75MHz?

Kai

Hi Shu-Cheng,

Are the parts actually failing, as in permanently damaged? Or missing pulses or irregular output?

The LM393 is not fast enough for the 6.75MHz signal.

Assuming the best-case response time of 300ns, and a rise/fall time of 100ns, the theoretical togggle rate is 1/(tphl+tplh+tr+tf) = 1/(300ns+300ns+100ns+100ns)=1/800ns = 1.25MHz. And even that number is optimistic, I would say < 1MHz.

For more info on this, please see the TI Precision Labs - Op Amps: Comparator Applications

So what you may be seeing as "failures" are devices that are just not fast enough and cannot keep up. The missing pulses in the output are a clue.

You need a faster comparator.

Working the above backwards, you need a total delay of about (1/6.8MHz) 147ns. So assuming 10ns rise/fall, you need a comparator with a max prop delay of 64ns. Of course, the faster the better.

Unfortunately, the 12V supply is a limiting factor for your choices. Most high-speed devices are 5V supply.

For 12V devices, you have the following choices:

 LM319 (50ns Dual)
 LMH7220 (5ns, LVDS Output)

For 5V devices

 TLV3202 (40ns)
 TLV3502 (4.5ms)

Of course, you will need to modify the reference point, too, if you change to 5V.

The TLV3202 could drop in replace the LM393 if the supply is dropped and the bias point needs to be brought to around 2.5V. The pull-up resistor can also be removed as these are push-pull outputs.

Hi Sirs,

Please refer our update as below.

The output always keep low on fail board, but the function will be normal after change IC.

Or change the R707 value larger than 845K.

We would like to know does LM393 have any tolerance will effect this phenomenon?

In addition, for adjusting the value of R707, can you confirm it?
This value needs to be set to exactly match the situation that this IC will not cause no output.

Thanks!!

Hi Sirs,

Sorry for pushed.

It’s very URGENT that we have 6% failure rate on factory and we need to stop the production.

What’s the different between the pass and fail IC since changing IC it can work.
And what’s the confidence value of the R707 for pass and fail IC since changing the R707 from 680K to 1M it can work.

Hi Shu-Cheng,

Paul already mentioned that the comparator is not fast enough. But there's also an issue with your transistor stages:

They don't toggle at all!

You could speed them up by the help of the following changes:

Of course, you could also use small signal MOSFETs to do the togglings.

shu1.TSC

Kai