Gareth,
Start off with a nice fat lot of caps for the pre-amp, keeping in mind your peak current consumption and the recharge rate then do a lot of unnecessary calculations to convince yourself that 2200uF is probably way too much.
Listen and hear if something differs from your expectations only then change it. Start simple and work your way to complex, complex is never the first option.
You cannot throw components at a problem, first establish if you have a problem. Keeping in mind that there is nothing on earth (semi conductor) that will dump current at the speed a capacitor can.
Don't believe in transistors multiplying capacitors, they cannot! It is a cheapskate solution if you cannot afford a few caps because transistors is a dime a dozen caps are not.
If you want to use a cheap series regulator so be it even an LM7815 works wonders to get rid of a little hum. If you need a little extra volts bias it with a zener to get to your 30V.
Years ago I thought I designed the "best amplifier" by zero compromises, I was not satisfied. Low and behold, I ended up trying an SMPS of dubious origins for a price I would think twice spending on a can of Pepsi. It fed a fat capacitor bank and it sounded magic.
I have not touched that system in six years and it remained my favorite till today. Although I designed and used others designers products since, nothing gave me that magic until I add the SMPS with a nice lot of caps and HF filtering with a little coil. You can purchase a nice SMPS 60V buck regulator for your pre-amp off eBay for less than two bucks (excuse the pun).
DC/DC LM2596 HV S 60V 3A Buck Constant Current/Voltage CC CV Step-Down Module | eBay
Last edited:
Member
Joined 2009
Paid Member
My hobby seems to be forever a moving target, I never really can be sure what my pre-amp will end up being used for ! But I know the phono needs to drive a volume pot so a buffer is helpful there. The output of the pre-amp needs to drive a power amp over some cables and it's possible that a low impedance volume control might be able to do that without an output buffer. I could indeed offer myself an output RCA that comes directly from the volume pot. and another RCA that goes via a buffer - have the choice of both without any real compromise. 🙂
Nico. Did you add a bit of resistance to the SMPS + caps? I would imagine even 0R1 would help ( RC filter ). If the caps were at the power amp even the joining cable might do plenty. I measured the resistance of a fuse noise wise once. 10 000 uF FUSE 10 000 uF, the fuse helped and was required somewhere in that circuit.
I go along with what you say about the PSU. If I were Gareth I would build the PSU as an upgradable feature. To keep myself honest I would try the LM317/337. Even 4 x PP3 batteries ( cheapest zinc one and 2 x 1 uF to help ). If it sounds better than the 30V PSU you will know not to throw money at it.
I use a cap multiplier ( 600V FET, 480 VDC out 140 mA, 5 V loss ) in a valve amp. It will beat anything and that includes chokes at a sensible price ( $15, FET + caps $3 ). My friend showed me how to float a LM317 to do the same. It was so complicated and looked trouble. The FET reduced from memory the hum from about - 67dB to -85/88dB ( 50/100 Hz ) reference 1 watt. For fun I substiuted a 60V darlington like MJ3001 that I had in the scrap box. A zener from emitter to collector as a crude version of the LM317 circuit. It protested a lot on start and then settled to work and did not fail. It was not able to match the FET although doing an OK job.
I would suggest no shunt regulator is really better than a LM317 or a FET multiplier for one very simple reason. All of them use the output cap to do the most important job of all. That is the > 1 MHz job. In truth the 0.1 uF or whatever at the gain stage is the one truely doing the job. Put your scope on the power rail if using either-net via mains. You will often see plenty. Takes my noise floor up about 6 dB. Nico you are right in most of what you say. Never ignore the multiplier, if an amplified zener make sure the filter cap does a good job below the zener knee voltage as a multiplier. A FET+ zener+cap+CCS and output cap is an interesting series device of low complexity. Although LM317 etc are very stable they are not totally so, not least if a pass transisor is added as a collector output as per data sheet for high current.
I go along with what you say about the PSU. If I were Gareth I would build the PSU as an upgradable feature. To keep myself honest I would try the LM317/337. Even 4 x PP3 batteries ( cheapest zinc one and 2 x 1 uF to help ). If it sounds better than the 30V PSU you will know not to throw money at it.
I use a cap multiplier ( 600V FET, 480 VDC out 140 mA, 5 V loss ) in a valve amp. It will beat anything and that includes chokes at a sensible price ( $15, FET + caps $3 ). My friend showed me how to float a LM317 to do the same. It was so complicated and looked trouble. The FET reduced from memory the hum from about - 67dB to -85/88dB ( 50/100 Hz ) reference 1 watt. For fun I substiuted a 60V darlington like MJ3001 that I had in the scrap box. A zener from emitter to collector as a crude version of the LM317 circuit. It protested a lot on start and then settled to work and did not fail. It was not able to match the FET although doing an OK job.
I would suggest no shunt regulator is really better than a LM317 or a FET multiplier for one very simple reason. All of them use the output cap to do the most important job of all. That is the > 1 MHz job. In truth the 0.1 uF or whatever at the gain stage is the one truely doing the job. Put your scope on the power rail if using either-net via mains. You will often see plenty. Takes my noise floor up about 6 dB. Nico you are right in most of what you say. Never ignore the multiplier, if an amplified zener make sure the filter cap does a good job below the zener knee voltage as a multiplier. A FET+ zener+cap+CCS and output cap is an interesting series device of low complexity. Although LM317 etc are very stable they are not totally so, not least if a pass transisor is added as a collector output as per data sheet for high current.
@Bigun,
I have already posted and mailed you improvements to your schematic in post #320, covering the phono stage, regulator and buffer. Nico has a lot of sage advice too...I would pay attention if performance is your priority. If building interesting looking circuits is your priority then carry on.
Wrt switch mode psus, I think this is a convenient alternative as long as you deal with noise. Dealing with noise usually means a separate box or plug mount and then you still need some local filtering to clean it up. Overall, seems less hassle just to use an off the shelf linear regulator.
I have already posted and mailed you improvements to your schematic in post #320, covering the phono stage, regulator and buffer. Nico has a lot of sage advice too...I would pay attention if performance is your priority. If building interesting looking circuits is your priority then carry on.
Wrt switch mode psus, I think this is a convenient alternative as long as you deal with noise. Dealing with noise usually means a separate box or plug mount and then you still need some local filtering to clean it up. Overall, seems less hassle just to use an off the shelf linear regulator.
Member
Joined 2009
Paid Member
The pcb will include shunt regulator, but it can be built without it. As you know, there will already be an upstream LM317/337 regulator in the box providing ripple-free +/-36V. I could probably tune that down to +/-30V and try without the shunt regulator by installing a pair of wire links on the pcb. The only limitation is time.
At >1MHz I'm wondering where the signal or noise at this frequency is coming from ? Are you thinking there will be diode rectifier noise coming down the cable from the external supply, through the upstream regulator, through the shunt regulator and inter-modulating / being detected by the sensitive front-end of the phono amp ? The rail caps I plan to use are Nichicon bipolar types with provision on the pcb for ceramic bypass. I've read that high speed bypass caps can cause more problems than they solve though.
Last edited:
Member
Joined 2009
Paid Member
Performance is a priority. And I'll listen to all the advice I can get. I may cherry pick from it. Let me see if I understand what items on my Boar#1 latest diagram are at odds with the advice so far:
a) Shunt Regulator:
i) don't need it at all and can rely on upstream regulator (can by bypassed)
ii) use a simpler shunt regulator (can be built without current mirror using existing pcb layout too if desired with higher Zout)
iii) stability concerns with Darlington device (can be built with regular BD139/BD140 if desired with higher Zout) - I'll add a zobel to the shunt output too just like any other negative feedback 'amplifier'
b) Buffer:
i) some folk don't like them - it too can be bypassed / not installed on the pcb but I think it'll be beneficial to the phono amp in driving a volume pot.
ii) an i.c. would be better - ? no argument here, but this is an example of where personal preference takes over. Hopefully my buffer is both simple (undamaging) and effective
c) Phono amp: - I've pretty much taken this from the NAD. I like the historical reference, the circuit simulates just great and has gobs of headroom. I'm not sure what advice I missed on this one regarding the design except
i) don't use a small cap on the input as it raises input impedance at L.F. which increases noise. I have doubled the value from 0.47uf to 1uF.
What ese did I miss ??
p.s. I had a look at the Blowtorch thread for some further advice. Here's what JC says about the power supply (bear in mind he's a JFET/FET focussed designer, my preamp is all-BJT and although I like FETs as much as BJTs it would be a start-over prospect to clone the Blowtorch instead):
Actually I may, at most, use 3 regulators in series. One is feedback and zener controlled to remove hum and line breathing, the second (if used) in order to remove high frequency glitches, and the third is always open loop and is actually part of the individual stage. It is usually designed to isolate that stage from other signals. It is simple, usually a cap follower, so it MUST have a stable DC input voltage that is supplied by the conventional high feedback zener referenced voltage regulator at the input.
The main factor that seems to be forgotten is that the regulators must remove both very low and very high frequencies . This is difficult with one regulator, at least, I don't know how to do it easily…the shunt regulator still can be very valuable because it can be much faster responding than the input series regulator, AND it can give a constant current load to the input series supply at all times. Not a bad thing at all.
Personally, I find the IC regulators marginal and I like to separate them from the actual working circuits. However, they DO reduce the hum and set a respectably constant reference voltage. To reduce transients, RFI, and audio noise, I use other circuits or passive parts either before or after the IC regulator. As far as design philosophy is concerned, I like to think about new ways of making circuits. Sometimes, I see a new way of looking at a circuit on this website. I find that interesting. However, why don't I like normal IC regulators that much? Well, they are noisy, have bad transient response, and fall apart at high frequencies. I prefer, myself, to use an active IC regulator in series with a simple high speed fet cap follower. This gives me the DC stability of the feedback regulator with the high speed response and low noise of the fet follower.
...as for the classical 3 terminal series regulators where sometimes a small resistor ~ 1R in series with the output lityc could improve the sound.
and on some other relevant topics he says:
I have used Alps pots in MANY preamps. TKD is better. Penny & Giles is also better.
Output impedance is fixed at 1000 ohms. I have always used transconductance amps for my line drives, when practical. They have several other advantages, as well as simplicity. Transconductance amps are more stable with a difficult load. I made the line driver for the Grateful Dead with this technique. Very difficult load, worked great!. This is because followers tend to ring due to the generation of -R due to phase shift at high frequencies. TA's just slow down, but don't ring very easily.
My designs are also simpler in thru-path than op amps, because they don't have the extra output stage.
I don't anticipate difficult loads so my thought is not to use transconductance line drivers but it would be simple to add a little series R to the Line-Out.
Last edited:
Member
Joined 2009
Paid Member
Got it 😀
Purely from a historical interest perspective, this also ensures that the -ve rail can be -25V instead of my -30V so operating points are the same.
Attachments
Last edited:
I tried simple cap-multiplier in power amplifier, and I do not like it. But I tried more complicated one (using LTP, current mirror, and buffer), I satisfied. Definitely, the mid and high quality is lot better. In cap-multiplier, do not ever make pass transistor saturated. Ever.
Actually it's just the hybrid pi small signal model of BJT operation, as taught at UC Berkeley, MIT, Caltech, et al. It's the model used in all analog circuit design papers published in IEEE journals, and in all post-1990 university level circuit design textbooks.
The hybrid pi model treats the BJT as a voltage controlled current source, plus some real but unwanted non-idealities. Such as: nonzero base current (noninfinite beta), nonzero output conductance (noninfinite Early voltage), and nonzero capacitance (noninfinite fT). Hybrid pi is wildly successful in practice and all analog ICs are designed with it, exclusively, today.
I would recommend extreme skepticism towards any circuit design suggestions made by anyone who disparages the hybrid pi model.
I know all about the hybrid pi model, thank you. It's a perfectly good model. I don't know what your education or experience or even age is, but I will offer you these words of wisdom: to make excellent circuits, not just good ones, it is very important to understand all the ways of looking at components and to understand how they work in practice, not just in textbooks. That way you can choose the correct model for the specific problem you are trying to solve. 🙂
I would recommend extreme skepticism towards any circuit design suggestions made by anyone who simply quotes things out of text books.😎
Last edited:
Member
Joined 2009
Paid Member
We're getting off track again - but we should remember the hybrid pi model was primarily a small-signal model. Good for pre-amps perhaps.
Member
Joined 2009
Paid Member
Ah, a fellow inventor !
It's been awhile for me. I published way back when I did my Phd and then had a bunch of patents but the past decade or so have pulled me away from technical things.
To some extent, I like this audio hobby because much of the technology has remained unchanged for about as long !
It's been awhile for me. I published way back when I did my Phd and then had a bunch of patents but the past decade or so have pulled me away from technical things.
To some extent, I like this audio hobby because much of the technology has remained unchanged for about as long !
Member
Joined 2009
Paid Member
Headphone amplifier - I'm not sure I remember how this got onto the agenda but it's been mentioned a few times. Thing is - I don't have a need for a headphone amplifier, this control-amp is for a speaker based system (I do own a pair of Bose noise cancelling headphones). So I'm checking back to my original goals and not seeing a good reason to include a headphone output - especially as it has significantly larger demands on current drive than a Line Out.
Why > 1 MHz. Here is the most overlooked reason I suspect and leads on from my JFET discussion. Some say JFET's sound better due to amplification curves. I have doubts about that. More likely the imformation below. What I wanted to bring into view is much folklaw about regulators may only concern in truth the output cap, decoupling cap and the device choice between bipolar and JFET. The track resistance being optimum also.
http://www.analog.com/media/en/training-seminars/tutorials/MT-096.pdf
http://www.analog.com/media/en/training-seminars/tutorials/MT-096.pdf
Member
Joined 2009
Paid Member
I agree, it's important to review the final design for EMI resistance and check that the right precautions are in place. I've found that it's quite rare to have an issue with local radio bleeding through to audio but I realize this is only what happens in the extreme case and we have to ensure that there is no low-level EMI impacting the background level as it may not be easily recognized through listening.
I was looking again at the NAC gain stage. It's not a bad design at all. With a singleton input it will tend to have higher 2nd than 3rd harmonic (which I view as desirable) and of course the whole thing operates in Class A with an output loaded on a CCS. Simulating it with 100% feedback i.e. as a buffer, it looks very good. Hmmmmm, if I were to squeeze this onto the phono board I'd have an all-in-one board for the active circuits.
If I put this all in front of a 1k, 5k or 10k ohm volume pot it could be quite simple. There would be no need for a buffer after the volume pot if it were low enough impedance ?
The simplest case becomes: source select -> (optional NAC gain/buffer stage) -> vol. pot -> line out
However, there are many subjective reports about improved sound from the use of an active buffer to drive Line Out so it is quite tempting to want to retain one.
I was looking again at the NAC gain stage. It's not a bad design at all. With a singleton input it will tend to have higher 2nd than 3rd harmonic (which I view as desirable) and of course the whole thing operates in Class A with an output loaded on a CCS. Simulating it with 100% feedback i.e. as a buffer, it looks very good. Hmmmmm, if I were to squeeze this onto the phono board I'd have an all-in-one board for the active circuits.
If I put this all in front of a 1k, 5k or 10k ohm volume pot it could be quite simple. There would be no need for a buffer after the volume pot if it were low enough impedance ?
The simplest case becomes: source select -> (optional NAC gain/buffer stage) -> vol. pot -> line out
However, there are many subjective reports about improved sound from the use of an active buffer to drive Line Out so it is quite tempting to want to retain one.
Last edited:
- Status
- Not open for further replies.