Snail Phono Pre

Mike H · Post by **Mike H** » Fri Oct 14, 2011 11:01 pm

No silly, CCS's for valve anodes.

Except the third one which is for a CF cathode. (If you want)

Note 200V supply, it's a bit of a clue.

The MOSFETS are 240V, 200mA, 750mW, E-line package.

jack · Post by **jack** » Sat Oct 15, 2011 4:02 am

Mike H wrote:Waitaminute waitaminute waitaminute wait wait wait[/img]

Fun. Theory is good for ideally modelled devices:
(3.3-1.4)/1300 = 1.46 mA

However, this assumes Vgs(off) in a real device is exactly -1.4V and the zener is exactly at 3.3V as per the models. A 2k2 trimmer for the resistor would work well as dissipation is only around 3mW. Mind you, the 10m45s has up to a 50% part-to-part variation and the schematic has no adjustment shown for them.

Having said that, I recently built an LTspice deck for the Quad 405-2 that actually behaves almost exactly like the real-world, or at least the same as the two modules I have to compare it with (except for noise/THD/thermal/PSRR behaviour etc, but that is down largely to the lack of sophistication in the models, true for all emulations).

Dave the bass · Post by **Dave the bass** » Sat Oct 15, 2011 7:44 am

Mike H wrote:No silly, CCS's for valve anodes.

Except the third one which is for a CF cathode. (If you want)

Note 200V supply, it's a bit of a clue.

The MOSFETS are 240V, 200mA, 750mW, E-line package.

Doh!, ta Mike. It was the 3.3v zeners that got me thinking.

DTB

Mike H · Post by **Mike H** » Sat Oct 15, 2011 1:03 pm

No prob.

nickds1 wrote:However, this assumes Vgs(off) in a real device is exactly -1.4V and the zener is exactly at 3.3V as per the models.

Natch, actually, datasheets say 0.7 to 2.0 Vgs, plus, the Zeners give 3.5V in sim, ergo I tweak the source resistor values in sim same as I'd have to tweak them in real life.

Though I don't like presets if I can help it

Mike H · Post by **Mike H** » Sat Oct 15, 2011 1:31 pm

nickds1 wrote:Having said that, I recently built an LTspice deck for the Quad 405-2 that actually behaves almost exactly like the real-world, or at least the same as the two modules I have to compare it with (except for noise/THD/thermal/PSRR behaviour etc, but that is down largely to the lack of sophistication in the models, true for all emulations).

There are some guys on the LTspice yahoo group get really bogged down in details, personally I adopted an approach to only use it to test the basic theory, real life still has tendencies for variations. I still think of it as like a paper drawing, just with the added advantage it can be test-run. The rest is up to vagaries of real world components. Same as, if you started with a paper drawing....

jack · Post by **jack** » Sat Oct 15, 2011 2:22 pm

Mike H wrote:There are some guys on the LTspice yahoo group get really bogged down in details, personally I adopted an approach to only use it to test the basic theory, real life still has tendencies for variations.

Agreed. There are bordering-clinical obsessives on that group - how you can realistically expect to model THD+N in a spice with general purpose models, I don't know. However, its an extremely helpful group and those times I needed a lot of help, e.g. with mixed-mode modelling, they've been very good indeed - it helps that the developer and lead support engineer are both very active on there.

Like you, I use LTspice as a scratchpad to test new ideas, then breadboard them before producing prototype PCBs - saves a huge amount of time, especially now that LTspice is multi-threaded (unlike a lot of SPICE engines). I like being able to play with ideas on my way up to the City, then test a probably "working" design that evening or weekend...

I will be posting a new very small project in the next few days that had its theory tested in LTspice before breadboarding on Stripboard (Vero) and then drawing up a PCB in Eagle. So simple to do...

Mike H · Post by **Mike H** » Sat Oct 15, 2011 2:42 pm

Oh yes didn't mean to imply the group isn't helpful cos it is.

Anyway enuff! Back to the Snail....

Mike H · Post by **Mike H** » Sat Oct 15, 2011 8:43 pm

Just noticed something, there's no resistor between input and ground?

little eddy · Post by **little eddy** » Sat Oct 15, 2011 10:22 pm

Mike H wrote:Just noticed something, there's no resistor between input and ground?

Sorry not on the drawing yet but there are to be 1:10 MC step up transformers and a load resistor to 0V.

Mike H · Post by **Mike H** » Sun Oct 16, 2011 12:09 pm

Oh right OK then.

little eddy · Post by **little eddy** » Tue Oct 18, 2011 11:07 pm

I think my calculator is giving up the ghost but I now have my first stab of the phono section with a single stage RIAA network.

Please note that for the resistor loaded option which you see, I could have got away with a much bigger initial RIAA capacitor of say 10nF but when I then moddelled this with a CCS (where I assumed 1M0), the series resistor fell to 126k, significantly below MJ's recommended 200k.

One big question is how can I find out or estimate the output impedance of the single and cascoded BSP129 and IXYZ 10m45s CCSs?

One final thing. Am I right in when calculating the series resistor, from the theoretical 317k, I work out the parallel resistance of the two input tubes, the load resistor and also the 2nd stage grid leak resistor? In this instance I get 21k4.

I have calculated/estimated the 2nd stage Miller/stray capacitance at 57pF.

Please feel free to point out the deliberate mistakes.

Post by **Nick** » Tue Oct 18, 2011 11:14 pm

All sounds about right. I cheat and use spice, but well worth at least working through your first one by hand.

Mike H · Post by **Mike H** » Wed Oct 19, 2011 1:52 pm

The ECC83's are going to have very low gain with just 27k anode resistor.

little eddy · Post by **little eddy** » Wed Oct 19, 2011 10:09 pm

Mike H wrote:The ECC83's are going to have very low gain with just 27k anode resistor.

I was trying to keep the RC filter hum attenuation as per the original builder because I can't recall any hum issues. Having larger currents, the stage voltages fall accordingly.

I could reduce the filter resistors to give say 250V in the hope that the hum doesn't increase significantly. I could always try to add more filter capacitance to offset the smaller resistors.

So I can redesign, perhaps based on the WAD Phono3 operating point of V+ 250V, Va 150, Ia 1.0mA and Vg of 1.1V. Using Av=u*Rl/(rp+Rl+Rk(u+1)), this will increase from 23 to 35 for the resistor-loaded version.

But when I switch to the CCS, if I understand this correctly, I just have to have sufficient volt drop across the CSS for it to operate correctly, so a supply voltage of 200V should be of no detriment. I would also have a similar power dissipation for the BSP129 as Nick's Phono clone at 0.1W but would this be an issue at 0.2W?

How do I work out the gain of the CCS version? If I assume an effecive Rout of the CCS at 1M then using the above equation the gain is 85, also as per a horizontal load line.

Mike H · Post by **Mike H** » Wed Oct 19, 2011 11:16 pm

LOL you made me resort to LTspice

(as well)

OK so with 1mA CCS ~

First problem is with no bypass caps on the cathodes gain is poor with EQ network connected. As, unbypassed cathodes increases output impedance. After I put 100uF on each I got a Voltage gain of 82.5 after the 0.1uF cap. I expected >=60-ish.

For a RIAA EQ freq. sweep it appears to be 2dB down at 20 Hz. Otherwise looks the right shape.

And indeed with a 100k anode resistor instead the gain is 65. RIAA curve is the same, so interestingly it appears not to matter whether a resistor or CCS is used.

PS 100k resistor gives about 115V DC at the anodes with a 220V supply, the CCS gives 110V.

You could drop the B+ for the CCS, quite likely will work and leave plenty of margin at 150V if that helps?

audio-talk

Snail Phono Pre

#106

#107

#108

#109

#110

#111

#112

#113

#114

#115

#116

#117

#118

#119

#120