Binary coded kind of log resistance

[warpie]

+1 here. Very neat solution!

 

[ruffrecords]

+1 here. Very neat solution!

It is. Some parts are 'clear. The third relay K3 selects between one of two resistor chains. The in each chain the other two relays select some combination of resistors to create the desired value.

Less clear (to me) is the two different arrangements of relays in each arm. The top network is relatively easy to understand. There are four resistors connected to two sets of relay contacts such that the series combination of either of the top ones with either of the bottom ones can be selected. The bottom network is not so simple. It very cleverly allows three resistors to be selected individually or all three in parallel. It took me a while with paper and pencil to convince myself the the 111 combination does actually connect all 3 resistors in parallel. It takes an excellent brain to create that. Much respect moamps.

Cheers

Ian

 

[Bo Deadly]

Yup. It would be interesting to see how many steps you could make with a 4th relay. Unfortunately it would not be able to leverage the 3 relay technique because the 2 states of K3 correspond to the 2 poles of each of the other relays. So it's limited to about 7 steps although the 7th step where all 3 resistors are in parallel is not as flexible so I think this technique is probably only really good for 6 steps. And solving for the values is a little difficult. An equation solver in a spreadsheet or JavaScipt calculator could make that easy though. If there were some sort of clever way to get 10 steps or more from 4 relays, that would be a significant improvement on the current limitations of digitally controlled analog gear (of which I'm partial to).

 

[zamproject]

what a clever network, chapeau !

Best
Zam

 

[moamps]

Thanks guys,
it is a specific solution for a specific problem.

This or similar approach can be used for 4 relays attenuator but the relays should be 3 or 4 pole variants which isn't simple anymore. In this type of application, I prefer using  reed relays or standard types (with bifurcated contacts). Cheaper reed relays are usually single pole so design mimics a standard switch.

Which type of control switch (or logic) MBB or BBM should be used, depends also of the position of resistors network, not only if it need to keep  DC path.  Here, for example, BBM should be used because in short time between steps the network goes to max. resistance and the system will go temporary to state of minimal gain which is good. If MBB type is used, in short times between steps, the network goes to lower resistance than adjacent steps are, so the system goes to higher gain in a short time which isn't good.  Sometimes a small delay in relay's switching the state caused with a protection diode can make a big problem.

There is a also situations where most part of an attenuator should be MBB and just one or two should be BBM ). One example is  the mic/line attenuator in NEVE preamps, and there is a famous OFF position where additional gain stage is inserted in a chain.