Why correct use of the optorelays is so bad?

After looking at the datasheet, it looks like the turn-on time leaves a lot to be desired as it is, at best, at least ten times slower than what you need.

It is true, but we do not use relays to generate signals. The frequency 40 kHz is not a switching frequency of devices, it is frequency of signals which go through.

Commutation by optorelays is made once during test procedure and after that relays change to operating mode which stays the same long time (hours).

In that case, no, I don't have any idea what's wrong. Are you operating at elevated temperatures? Are currents at near maximum?

Temperature was normal. It was a prototype, which was investigated on the table in laboratory and dissipating energy was negligible. I have mentioned in my first post that all parameters specified in datasheet are 10% of maximum values.

Are you switching inductive loads which may cause back EMFs which exceed the device rating?

Yes, there are transformers on both transmitting and recieving parts and the relays are commutating these Tx/Rx circuits to make internal loopback and to disconnect external contacts during test.

We had the same guess about reactive EMF in process of commutation, so we mesured instant voltage pulses during commutation with wide band oscilloscope. Nothing special was observed. Peak pulses at the moment of switch had amplitudes aprox. 3 times more than regular (3V vs 1V). Taking into account that manufacturer boasts of 60 V minimum for blocking voltage it does not explain what we have.

Post the circuit please.

Could you please actually give the values that are "only 10%" as with the drive current only being 1ma up to a max of 3ma if above 60°C, 10% will be far too little....

We need some full accurate infos and a circuit diagram.....

Here are real values of CPC1018 in the electric curcuit vs maximum ratings given by manufacurer: - input control current 7 or 10 mA (50 mA); - load current 20 mA (600 mA); - blocking voltage 1 V, peak 3 V (60 V).

It is interesting to know that CPC1016N do not degrade in the same curcuitry, but we are not satisfied with their On-resistance. I think, that difference in blocking voltage (100 V vs 60 V) cannot explain the degradation of CPC1018 if voltage of only 3 V is really measured.

Input control current should be around 1 ma with a max of 3 ma if in warm conditions (above 60°C I think it said), a higher value of series resistor is needed...

Although it states a max of 50ma, and you are well under that, try reducing it to under 1ma and see if that helps....

Here you can see a part of circuit which gives the idea of commutation environmment. The relay K71 is switching on termination (load) resistor R169 when internal loopback operates. Tansmitted signal is generated by transistor VT16 in key mode. Frequency of control sequence KM_TX2 can variate up to 40 KHz. There are other relays to commutate inernal loopback or regular operation configuration, which are not shown. K71 is first degrading, we tryed diod restriction, shown red (though we didn't see the necessity on oscilloscope) and it didn't help. Relay control signal KM_CTR2 changes once only during selftest procedure.

The quality is so poor, I cannot read what components are what. I will send you my email address via CR4 email...please send a better copy there, thanks.

Change R49 to 3K ohms and see if that fixes the problem. I feel you are damaging the LED in some way.....

You may want to remove eventually (if the 3Kohms fixes the problem) the two extra diodes from the "contact" side later too, I do not see how they can help, but I cannot see the signal levels either at that point.....they will remove and reverse voltage, but spikes in the normal direction will still get through. If there is a signal problem at that point, you may want to connect two pairs of silicon diodes reversed to clamp any voltages in either direction to 0.6 volts....

As in:-

It helps to clamp in both directions!! You may need to add a series resistor of low value to current limit to diodes depending upon circuit impedance at that point.