Best way to install Superreg in a preamp

[hacker]

I've ordered some PCBs from Andy and am awaiting the parts from Farnell (which cost about £20 less than RS), but when I finally come to hook up the superregs, I'm curious as to the best way to do it.

Current setup: 62/140. The 140 is powered from the mains by standard Naim mains cord. The 62 is powered from the 140 by standard Naim 4-pin Snaic. The 62 has the buffer stage bypassed, to there's only the gain stages to power. The power rail decoupling is done by 56u/25v oscons bypassed with 680pf Wima green caps.

I've measured the DC across the pins of my 62's 4-pin connector and I've got 24.4v coming into it from the 140(not much headroom for my oscons!)

There are a few ways to tackle this and I'd appreciate feedback:

1. Mount SR in the 62 and configure it to output 19.4v from the 24.4v provided by the 140. This is the easiest solution, but will running the gain stage at 19.4v be sufficient? Neil McBride's website suggests it will be, but anyone else like to offer insight?

2. Modify the output of the 140 to provide 30v, then do the same as above except configure the SR to output 24.4 as it's doing just now. Easy? Hard? Sensible? Possible?

3. Mount the SR in the 140, replacing whatever regulation is already present. This presents a few challenges (I'm not even sure what they are, considering I don't know the layout of the 140 yet - this is all theory). Is it possible or even sensible?

4. Any other ideas? ("buy a hicap/snaps" is not a useful suggestion )

Cheers,
Carl

 

[martin clark]

1) Yes, but aim for 18-18.5v in case the supply rails /mains supply sags. 18v is plenty, though more seems to open up the sound slightly. Quality is more important than quantity...that said:
2) If you can increase the voltage from the 140, do try it. See 1 above, I'd suggest you leave 6v headroom for the SR - it probably only needs 4.5v, but speaker load and mains supply variances will drag the 140's rails down.
3) no point - space is tight and the regs are better off closer to the circuit they serve.
4) An independant supply will be better, so build your own at low cost: http://www.acoustica.org.uk/t/naim/powersupplies.html

 

[JimmyB]

Not used Andy's regulators (used peranders SF) but instead of using the feed from the NAP140 why not build a basic power supply in a seperate box & feed the SP's mounted in the 62. This would allow you a lot more versatility & also avoid the leeching from the 140 (the classic mod for a 140 is to remove the preamp supply altogether, so I guess increasing the voltage from the 140 to feed the SP probably isn't ideal).

Have fun.

 

[Mus]

1) Yes, but aim for 18-18.5v in case the supply rails /mains supply sags. 18v is plenty...

The mute relay switching circuit needs >24V, unless it's been removed?

Regards,

Mus

 

[trancera]

I was asking Andy about exactly the same thing. Im on 140 / 62 too.

He suggested that biggest benefit would likely be had by bypassing the regulation for pre amps in the 140 all together.

Not sure if this means after a PSU cap or what. I'd suspect so unless you can fit a high quality smoothing cap in the pre. (probably cos 62s have loads of space). The only problem with that i see is putting a rough as flick DC on the same cable as the signal.

Anyway doing that would certainly have a plenty high voltage.

I'll be following this thread closely.



Incidently what was your total cost of two boards + parts ?
(I need to be price / performance conscious too )

 

[martin clark]

Mus, you're quite right, but that's supplied from the raw supply rail. In which case the 'in' before the Sreg should still operate the muting relay.

Easy thing to hack into shape if I've got this wrong though

 

[Mus]

Easy thing to hack into shape if I've got this wrong though

Nope, you're spot on. It will sound better with the relay removed though.

Regards,

Mus

 

[hacker]

The mute relay switching circuit needs >24V, unless it's been removed?

It hasn't been removed and I'm wary about doing so. Are there likely to be problems when I switch on/off the 140? (eg. thumps, bangs and crackles)

[Andrew Weekes] suggested that biggest benefit would likely be had by bypassing the regulation for pre amps in the 140 all together.

Hmm, cool idea. I wonder what voltage the 140 is spitting out pre-regulation? If its 30v - 36v then it's probably a winner. I could run that to the 62, feed it into an SR then the SR into the PCB on the 62. No need to chop out the relay.

Mus, you're quite right, but that's supplied from the raw supply rail. In which case the 'in' before the Sreg should still operate the muting relay.

So basically output from 140 into the 62's relay stage, then the relay stage into the SR, then the SR into the gain sections?

1. Yes, but aim for 18-18.5v in case the supply rails /mains supply sags. 18v is plenty, though more seems to open up the sound slightly.

Andy has provided values for R8 and R9 that would give me 19v from my 24.4v supply from the 140. The question is, what about the relay? Will bypassing it have adverse effects? Can I power it with only 19v?

Cheers,
thanks guys
Carl

 

[bivalve]

Carl,

I measured the voltage at which the relay board from my 32 turns on. It is 21.5V.
If you don't want to go mucking about with your 140 you might be able to figure out how to alter the 62's relay threshold voltage.

I would keep the relay with the 62 powered from the 140. however, I was surprised to find my multimeter measured 200kohm across each channel. subjectively I found removing the relay board to be less of an effect than removing the line amps.

Two Superregs is the way to go of course (recent convert!), then an independent power supply to drive them.

Regards,
David

 

[Basil]

One thing to bear in mind if you do bypass the relay board.

If there is a power cut whilst you’re out, the power amp and speakers will be fed the full output of the pre as it stabilises when power is restored.

 

[chiily]

Carl,

I had the same sort of dilema here

I found 18.6v not enough.

I ended up with six SRs in a separate box, running the six gnd senses, six gnd and six power lines into my 32.5 all providing around 24v to separate cards.

 

[bivalve]

Basil,

I agree with you there.

Carl,

Maybe the 62 layout will let you power the relay circuitry with the direct feed from the 140, and you can isolate the power connection to the gain sections and connect the output of the superreg(s) direct to them, maybe cut one end of the 27R resistors and air wire them.

David

 

[hacker]

Maybe the 62 layout will let you power the relay circuitry with the direct feed from the 140, and you can isolate the power connection to the gain sections and connect the output of the superreg(s) direct to them, maybe cut one end of the 27R resistors and air wire them.

Ok, I've checked it out and the relay IC is a PRMA2A21, docs for which are here .

Specs are:
Coil voltage: 12(typical) 21(max)
Operate voltage: 1(min) 9(max)
Release voltage: 1(min) 9(max)

All the operating voltages are under 24.4v (my input DCV) so must be controlled by resistors... if I adjust the R values, I should be able to adjust the point at which the relay switches. This means a lot of pissing about, not to mention adjusting the internals of my 140, too.

Much simpler is to take the 24.4v input of the 62 and split it in two - one feeds into the relay and the other into the SR. This makes the relay function as normal and the SR will be configured to give an output of 19v (with 5.4v headroom) to feed the gain stages.

This means that I don't have to remove the relay circuit (which appears to be a bad idea without a dedicated preamp PSU) and I have a much simpler job in the 62. The only question is the impact of running the gain stage at 19v as opposed to 24.4v.

If that sounds shit, then I'll have to assume that 19v isn't enough and I'll bite the bullet and mod the 140 to provide 32/33v to the 62. That can then be regulated to 24.4v by the SR with lots of headroom for voltage fluctuations in the supply from the 140.

Sounds simple in theory... I love it when a plan comes together!

Carl

 

[martin clark]

Much simpler is to take the 24.4v input of the 62 and split it in two - one feeds into the relay and the other into the SR. This makes the relay function as normal

That's exactly what I meant in post #6

 

[bivalve]

Me too.

You might consider bypassing the tracking preregulator (TPR). The TPR forces the superreg to drop ~2.55V between the output of the three terminal regulator and the output of the opamp section. I use red LEDs, green LEDs ar around .5V higher. In addition a LT1085/6 as the TPR should have around 1.5V* across it as well. Walt Jung designed the original regulator as a low dropout circuit (around 0.9V IIRC), and its only the introduction of the TPR that has lead to dropout voltages >4.5V.

Walt tested the opamp-only regulator and found it to have near the theoretical noise floor and microohm output impedance up to around 100KHz. He addred the TPR to improve line rejection. The TPR also limits the heat generated in the pass transistor to a small fixed amount, but the regulator in the 140 is doing most of this for you. I don't know what the difference in sound quality between TPR and non-TPR approaches might be, but the original non-TPR regulator was a radical improvement regulators.

* The absolute minimum is supposed to be better than 1V but this is at the bottom of any ripple trough imposed on your DC, allowing for mains sag and the effects of other lads on the supply.

David

(see next post)

 

[martin clark]

Of course the more elegant solution is an offboard DIY 30v supply

 

[hacker]

You might consider bypassing the tracking preregulator (TPR). The TPR forces the superreg to drop ~2.55V between the output of the three terminal regulator and the output of the opamp section. A LT1085/6 as the TPR must have around 1.5V across it as well. Walt Jung designed the original regulator as a low dropout circuit (around 0.9V IIRC), and its only the TPR that has lead to dropout voltages >4.5V.

Hmmm, interesting. So what headroom would you suggest I use if I bypass the TPR? If Walt reckons 0.9v, then perhaps if I take into account the needs of the 140, 3v may be enough? I want to keep the voltage in the 62 as high as possible and if I can get away with a 3v voltage drop, that would give me 21.4v to run my pre.

[update] Ok, I've just read the SR manual (duh, maybe should have done that first) and it says that input -> output voltage difference must be at least 2.5v without TPR and at least 5v with TPR. So, if I aim for 21v in my 62, that leaves a decent 3.4v headroom. Phew!

Looking good for a marathon soldering session this weekend. I've got my silver-coated OFC for rewiring the NAC; I've got the components for my SR; the SR boards are due to arrive this morning... I'll do a photo-shoot as usual and put them up at http://haggis.kicks-ass.net/audio/ for you guys to see.

Incidently what was your total cost of two boards + parts ?

£5 per board (I ordered 5) and approx £30 for parts from Farnell, although that's enough to cover most of the components for lots of SRs. It'll probably cost another £7-10 per board for stuff like the opamp, regulator, reference etc. So it costs about £35 for one board if you DIY; it'll cost £62 for Andy to build you one. You can also get him to provide boards with the opamp (AD825 I think) pre-installed for £12 per board.

Carl

 

[bivalve]

Today I measured the dropout voltage of a 15.185V SR, bypassing the TPR, and got the following:

drawing 21.40mA 0.767V
drawing 40.15mA 0.788V

The threshold was when my multimeter decreased by one count (10 microamp). During the test the regulator maintained its output voltage to within 1mV with an input voltage from 30V down to <16V.

However the red LED appeared to be extinguished at this point, but increasing the I/O voltage back to 1.5V turned it on to about the intensity of a full SR.

The LED intensity increases with input voltage. LEDs have an effective resistance (change in voltage with current) , so ripple or other voltage fluctuations will be passed into the current source output. One way to improve things is to replace R4 by a current source diode, and I have bought J511 (from RS) for this.

A red LED's incremental resistance decreases with current up to its maximum current. The ones I use (TLUR4400, MV50640 from Farnell) have a knee at around 5mA, below which their resistance is high, and above which the resistance is more constant. The J511 (the largest current constant current diode in this series, I think) produces around 4.6mA, so should do nicely. I have bought some but havn't tried 'em yet.

A green LED has worse resistance characteristics.

I'm trying to attach an XL chart; any tips. Meanwhile my little experiment is still on the table glowing quietly.

Regards,
David

 

[bivalve]

Carl,

Two more points:
1. From my two readings it seems that dropout voltage increases with current. A single gain board draws 10mA DC. I don't think it will draw much more with a signal, so 2.5V drop looks like plenty of margin, but experimenting may get you another 0.5-1V.
2. Of the 13 +ve SR boards I received from Andy, 4 had a short between the R8 land nearest the +ve sense connection point and the ground plane. It is very small, and in the 9 o'clock - 12 o'clock quadrant of the hole. Testing the resistance between the two areas is the best way to check for the short.

David

 

[peranders]

A red LED's incremental resistance decreases with current up to its maximum current. The ones I use (TLUR4400, MV50640 from Farnell) have a knee at around 5mA, below which their resistance is high, and above which the resistance is more constant. The J511 (the largest current constant current diode in this series, I think) produces around 4.6mA, so should do nicely. I have bought some but havn't tried 'em yet.

Using a preregulator is a good way to make the influence of the LED less.