advertisement


Replacing 4066 switches?

JimmyB

JimmyB

pfm Member
To avoid further pollution of another thread, I thought I'd ask if anyone knew of a circuit or ready made PCB to replace 4066's in the NAC112x with relays?

The Quads also had these and it looks like Dada maybe did something but can't find it on their website now. What remains is here:

http://quadrevisionspot.blogspot.com/2010/12/best-cd4066-for-quad-44-is.html?m=1

Not sure what the switching voltage used in the Naim units is but since the 4066 takes up to 15v, I'd hope there was enough to drive a circuit to switch relays.
 
As it was a comment by me that started this off on the other thread I though I'd better try to help if I can.

It's not a relay circuit but have a look at the ANT4066. It's expensive mind you if it's still available.
http://www.ant-audio.co.uk/index.php?cat=post&qry=ant4066
The datasheet does give some interesting information about various versions of the 4066 with the HC version coming out best.
http://www.ant-audio.co.uk/4066/ANT4066_datasheet_1_1.pdf

Maybe if you have one input you use for serious listening then you could put an ANT4066 there and replace the others with HC4066

Having looked at a few HC4066 datasheets, there's some variability in the 'on' resistance of those as well. Lower resistance is correlated with less variation which results in lower distortion. (Edited from "Lower resistance results in lower distortion.")
Toshiba chips used by Naim: ~100ohms
Philips 74HC4066 ~50ohms
Texas CD77HC4066 ~15ohms

So if you can find them, just replacing the Toshibas with Texas Instruments CD74HC4066 would bring a worthwhile decrease in distortion.

Do you use more than one input on the amp? Only asking because I only ever use the one input that's connected to my dac. If you only use one then an off-the-wall suggestion would be to just hard wire that input bypassing the 4066s completely.
 
Last edited:
The TI 74HC4066 is 10.5V absolute maximum
There are plenty of much better switch ICs, but all in 16 pin, not 14 pin packages
 
JensenHealey said:
Why would a resistance increase distortion?
Click to expand...
Werner said:
Because it is non-linear?
Click to expand...

Yes, the 'on' resistance is variable dependant on the voltage of the signal at that point.
If the resistance was constant, it wouldn't matter what the 'on' resistance was. Lower is better in this situation because this is correlated with less variation over the signal. HC4066 measured variation of 16->18 ohms, TC4066 variation measured between 85 and 108 ohms.
If it's ok with you all I'll make that more explicit in the original post, thanks.

davidsrsb said:
The TI 74HC4066 is 10.5V absolute maximum
Click to expand...
Good spot - I don't know what the exact implementation (inc voltages) that Naim used so I'd have to go through the datasheets to see if there was one more suitable.
If you're going for a different package you may as well shift to relays I would have thought.
 
DaveMc64 said:
Yes, the 'on' resistance is variable dependant on the voltage of the signal at that point.
If the resistance was constant, it wouldn't matter what the 'on' resistance was. Lower is better in this situation because this is correlated with less variation over the signal. HC4066 measured variation of 16->18 ohms, TC4066 variation measured between 85 and 108 ohms.
If it's ok with you all I'll make that more explicit in the original post, thanks.


Good spot - I don't know what the exact implementation (inc voltages) that Naim used so I'd have to go through the datasheets to see if there was one more suitable.
If you're going for a different package you may as well shift to relays I would have thought.
Click to expand...
Thx
 
Very similar 74HC4066 part numbers have absolute maximum all over the place. You have to be very careful, the Quad 44 is pushing to the limit with rather poorly regulated +/- 7.5 V.
The switch driver ICs seem to be unobtanium.
 
You could make a little PCB with a 16 pin SMD switch IC and pins to the 14 pin DIL footprint
Something like a DG211
 
Look, the Naim of old (maybe not relevant to this design though) would audition components in critical paths and buy the one they liked the sound of. Yes there may be better specced 4066s but I bet they listened to several variants and settled on the one that's actually fitted.
 
Yes, I'm a bit dubious that the 112x get a poor rep because of the input switching. Plenty of 4066 in use in studio equipment alongside things like NE5532's. My suspicion is that there are more factors in play but people are suspicious of silicon devices and they're the first to be blamed, even if the actual culprit is sub-ideal design.
IIRC, Rega used a Wolfson chip for the Cursa 3 volume control and it's better than a scratchy carbon loaded track IMO.
I may take a single input and put it direct through when I get to it...just to test the idea.
If we could work out what was so bad about the NAC112x, it could make a great cheap basis for a ready made remote control chassis.
 
For some unknown reason, the premium switch parts are not pin compatible with the 4066. Back in the design days of these products, the fancy switches were expensive.
I don't see them being a major problem as the distortion goes down towards zero at low levels - this is not crossover distortion.
 
MAX4066 and there's a thread about it on Tapeheads cassette forum. An example of expectation bias and a few persons arguing against the need to replace most 4066s.
I once replaced ALL 4066s in a Quad 34 with MAX4066 and heard no difference. Shorting out some elcos in the signal path did...
 
The MAX4066 looks good, very linear at 15V supply, available from RS.
You would have to listening with electrostatics to stand a chance of hearing even the CD4066 levels though.
 
Zombie, well done. "I did these mods: not a toss of difference!"
Somewhere else somebody suggested a before and after distortion measurement with the Maxim4066s (on a Quad34). Seems fine for someone with far too much disposable time.
If you were doing the whole thing now, yes, probably use the Maxim ones, or more likely the Vishay DG series (different architecture, not interchangeable).
On the other hand, I worked for a console manufacturer who used shedloads of 4000 series switches/muxes for their router/switchers so price was a factor. It's how you use them that makes them work; switching current at virtual earth point. See Douglas Self articles on this as well.
 
Ampex185 said:
It's how you use them that makes them work; switching current at virtual earth point. See Douglas Self articles on this as well.
Click to expand...
Using them on preamp output as voltage switches and where the switch load is not high enough leads to the distortion measured.
The audibility is questionable. The Maxim 4066 manages a fairly stable resistance with voltage, so it will measure far better
 
Did some research on the 4066-like chip (with LED drivers) in my Nak CD1:
"The TC9145P output has a 250 kOhm load and that is the point where the distorsion curves flatten out. THD at 20 kHz is 0,004 % and at 1 kHz and 100 Hz it is 0,0025 %. This given that the signal from the preceeding stage is not exceeding 1 V."
There is also a switching circuit in the DR1 (same circuit as the CD1) that uses FETs instead...what was the reason for the change? The chip was phased out from production or better sound?
 
You must log in or register to reply here.


advertisement


Back
Top