New Posts New Posts RSS Feed - Phono Preamp Pt2: MC
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Welcome to the Graham Slee Audio Products Owners Club

 

Open to all owners plus those contemplating the purchase of a Graham Slee audio product wishing to use our loaner program: join here
Subscribe to our newsletter here (Rules on posting can be found here)


Phono Preamp Pt2: MC

 Post Reply Post Reply Page  <12345 10>
Author
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8559
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 06 Jan 2013 at 8:59pm
Just getting back up to speed with all this because I think we need some new DIY on here to get people interested all over again.

So it's back to op-amps and some reasoning behind the type we need to use.

A refresher here wouldn't go amiss. The output of a low output moving coil phono cartridge is in the range 0.1 to 1mV - usually around 0.3 - 0.5mV.

That's where the output is rated at 5cm/sec which is the groove velocity, and it is stated at 1kHz which is a reference frequency because the whole frequency response is tilted because of the "alternator" type output of magnetic cartridges.

We also know that there is +14dB of headroom or overload margin in the "grooves" so the output can rise by 5 times (5 times = 14dB).

So, the largest output we have to contend with is 5mV. But wait a minute! The response is climbing which means it's going to be 20dB more at around 20kHz, and 20dB is 10 times more... 10 times more than 5mV, and that's easy - it's 50mV.

Taking this to its logical conclusion we need to understand that the output is in r.m.s. so the peak single sided output is going to be around 1.4 times bigger, and peak to peak that's 2.8 times bigger, leading to an output voltage of 140mV (yes, it's possible).

However, it's doubtful that any record mastering engineer would allow +14dB maximum excursions at or near 20kHz simply because most styli couldn't track that high, and maybe the cutting head would do a tizzy too?

See Kevin Gray Great Sounding Records for more on this.

Why am I going on about this? Well it's because in a non-inverting amp (discrete or op-amp doesn't matter) there is only 60mV of "give" at the input before the first transistor is saturated. Once saturated the next stage, which is an integrator stage, takes up where the input transistor leaves off and not only does it have to drive the integrator capacitor but the much lower impedance voltage amp stage (VAS).

So that's how close an MC stage is to disaster. You can't use FET inputs because they're too noisy at this level, or valves because they're like FETs. You can use a transformer if you don't want clean bass.

It would help if we could choose an op-amp that has some emitter degeneration resistors for its input transistors. The problem here is those resistors make things more noisy too. If people were able to put up with a little more hiss than normal, then that would be OK, and there would be more slew rate.

On the subject of slew rate it's worked out like this: SR = 0.3 x Ft. Ft is the transition frequency of the op-amp where it goes from gain to no gain - in other words a gain of unity which is 1, and 1 is a bit too low for what we need.

So with a natural 5 to 10MHz op-amp the slew rate is going to be 1.5 to 3V per micro second. That's going to sound as boring as many commercial MC amps...

If there were emitter degeneration resistors the formula gets rewritten as SR = 0.3m x Ft. You'll see the additional term "m" which is a multiplier higher than 1. If the emitter degeneration resistors were the same as the intrinsic emitter resistances of the input transistors m would equal 2, and that would double the slew rate. 4 would be better.

As I mentioned before the NE5532 is believed to have some emitter degeneration but try as I might I can find nobody left at Signetics to vouch for that. It does however manage 9V/uS which is 1.5 times better slew rate than the single version NE5534, when suitably compensated.

The problem with the NE5532 (and NE5534) is its asymmetrical output stage which has trouble charging and discharging the RIAA EQ capacitors.

It's a minefield really.

It can be done discretely and the AES (Audio Engineering Society) has suggested (many years ago) a suitable circuit, but that was for moving magnet. We need 10 times more gain than this.

OK, time for break. Hope you followed the reasoning so far?
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8559
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 06 Jan 2013 at 9:53pm
Maybe I got it wrong about the NE5532AN. Seems to be working fine here, but it needs to go on the audio analyser to check out RIAA accuracy.

Maybe I got the balance right by accident? The maximum capacitive load it can drive (forgetting stability for the time being) is I/SR which is rated output current over slew rate. I get 16(mA)/9(V/uS) = 1.78nF. Therefore the dominant RIAA EQ capacitor can be around 2.2nF.

I'll do some checking.
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8559
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jul 2013 at 3:37am
Six months further down the line and most of you will be thinking this project is going nowhere.

There's a good reason for that. I spent many years developing the two MC stages I do commercially: the Reflex C and the Gram Amp 3 Fanfare. They're both good and there's a reason why: they use the only two viable op-amps this earth has ever been blessed with. No other op-amp ever devised this side of "Alpha Centauri" can do what those op-amps do!

There is no way I'm going to make public these commercial designs - a chap has to make a living and that's final.

I promised you an MC preamp and I'm going to deliver one, but it's now not going to be an all in one.

We're going to have to go the same way as the Elevator EXP - it has to be a step-up amp which can be used in front of the Genera.

At this point I'm going to introduce you to a great DIY audio designer called Rod Elliot of Elliot Sound Products from downunder in Sydney, Australia: see http://sound.westhost.com/

I've never met Rod or even communicated with him, but his circuits are OK and he's good at explaining them.

I don't like any of his class A power amp designs but not because I think they're bad - in fact the total opposite - it's just that I consider class A to be mostly BS and unworkable in the real world. I've worked on class A power amps and they are notoriously unstable (which kills speakers), and in most class A power amp designs, they're really not class A at all, but class B with a bit more bias. Ask members like Leo (also one of our sub-contractors) who has also had to work on class A power amps and he'll vouch for what I have to say.

Class A belongs to preamplification and this circuit here is as good a starting place for the MC pre-pre as any: http://sound.westhost.com/project13.htm

OK, it's a mic preamp, but it's a low impedance mic preamp and MC cartridges are low impedance.

I'm not saying this preamp will do exactly what you want is it is but it will do something. If you feel like trying it, give Rod the cash required for the board and get building. What I'm going to do is a similar circuit and see if it can do the same job as an Elevator EXP. If this works out, Rod's PCB may be worth purchasing for those in easy reach of Sydney from a postal services point of view.

The circuit is a high output impedance transistor voltage amp followed by a high input impedance buffer transistor. Both transistors are the same type - it's the way they're used.

The second transistor supplies a constant current load to the collector of the first transistor by means of R2,3 and C2. It's called a bootstrap.

For this to work in conjunction with a Genera it needs a "local" voltage regulator and power to that from the Genera power supply. Obviously the power supply voltage won't be the 30V or the 12V shown in Rod's article. It will be 18V. Therefore some mods to component values will be needed.

I also think it should work just as well using BC337 transistors. The supply of good low noise transistors in this country has just about dried up and I suspect it's the same world-wide. The market for great transistors like the BC109C seems to have expired many years ago and the price of such beasties is now exorbitant. There's also the fact that they don't do the guaranteed high current gain C version anymore. That's progress for you. Also, using the BC109C was not really easy because of its low voltage an current capabilities which has caught out many a DIYer.

Right! Over the intervening period, however long that may be to my next post here, I'll be sticking the circuit on a piece of strip board and toying with values and seeing what I can come up with. If it sounds good and measures good I'll try and package it so as others can make it.

Before closing I just need to say that doing a pre-pre for moving coil has always been the most sensible idea. As it was at the beginning so shall it be at the end.

Graham
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8559
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jul 2013 at 9:38am
Before starting to assemble and test the circuit I thought it prudent to simulate it.

My suspicion was that the phase and gain margins could be a little too slim - something discrete designs seem to suffer from quite often. It leads to a rather glassy sound after some hours of use. And anyway, we don't want an oscillator.

By including a small ceramic capacitor in the negative feedback from Q2 emitter to Q1 base, value 47pf, we get nearly 90 degrees phase margin and 20dB (x10) gain margin.

The overall negative feedback is taken from after C3 to ensure no DC on the input. The simulation saw peaking in the very low bass, which is not a good idea. By increasing the value of C3 to 100uF the peaking is gone.

All the above is based on a 30V supply. Later I will simulate it on the required 18V supply and tweak it further.
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8559
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jul 2013 at 1:58pm
So here we are with the plot from the 18V version of Rod's circuit suitably tweaked about.



I guess by now it can't be classed as a Rod Elliot design but as he's the inspiration behind it I feel he must take some credit.

The broken line is the frequency response plot showing a gain of about 23dB (same as the Elevator EXP), and it's flat from 1Hz to around 100kHz (-3dB or half power points).

The line with "blocks" in it shows the open loop response and demonstrates there's 40dB of negative feedback so the distortion should be appreciably low. At 10kHz you'll see it's fallen to around 20dB loop gain and so distortion will be rising from there, but still it should be pretty low.

The other line is phase (the plain line). You will see I've placed the cursors at unity gain and at -360 degrees. As this stage inverts then -360 degrees is the equivalent of -180 degrees on a non-inverting stage - the point where a circuit oscillates. You can see it has around 90 degrees phase margin.

Where the -360 degrees cursor crosses the gain curve it can be seen it's -20dB which is its gain margin.

So far so good.

Next up will be the simulation schematic.


Edited by Graham Slee - 15 Jul 2013 at 1:58pm
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8559
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jul 2013 at 2:09pm
Here's the schematic...



More about it later...

School run time!
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 8559
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jul 2013 at 4:40pm
The circuit comprises a DC coupled transistor pair within multiple negative feedback loops. You've probably heard that negative feedback is bad so therefore I'm using plenty of it here! Misinformation is great! You just need to know it's misinformation and go the opposite direction like I always do...

The input represented by V1 (a simulator AC voltage source) is paralleled by the input load resistor R9 (120 Ohms) and R4 (1.2k Ohms) which, at its junction with R1 (18k) is a "virtual earth" or summing junction. Therefore the input impedance (which is purely resistive) is 1.2k||120R = 109 Ohms. This is close enough to the regular 100 Ohm loading most MC cartridges require. If you want to go higher then increase R9 but at some point R4 will dominate and therefore the highest load resistance is 1.2k.

The stage is inverting. The voltage gain is R1/(R4 + source impedance). As the source impedance is going to be in the order of only a few Ohms we can simply call it 18k/1.2k = 15 (23.5dB).

Bias to Q1 base is derived from the resistor split R6/R3 via R5. The voltage here will be around 0.6V. This causes Q1 collector to sit at around half power supply rail voltage (~9V). Therefore Q2 base will be the same DC voltage as Q1 collector, and the output (before DC blocking capacitor C3) will be around 9 volts. The signal can then swing symmetrically to within around a volt of each rail. It should do around 5.5V rms max. - plenty of headroom.

Q2 emitter load R6 and R3 in series is where the output voltage is developed across. The divider it forms and the value 100k for R5 gives a small amount of negative feedback to Q1 base and calculations suggest it will result in 60dB or thereabouts open loop gain - which the graph confirms.

Q1 is run at around 200uA which Rod Elliot found to be within the curves of constant noise from the data for low noise transistors, and I did too. The area within the curves of constant noise is chosen for the amplifier source resistance which is virtually R4 (1.2k) and at a frequency of 1kHz. At lower frequencies the noise will rise but all devices do similarly.

R2 and R7 form with C2 a bootstrap circuit which does the job of constant current supply to Q1 collector. It also has the effect of raising the input impedance of Q2. Therefore the gain of Q1 is high and it isn't unduly loaded by the input of Q2 (Q2 base).

All devices have capacitance which prevents them from being amplifiers at very high frequencies - they eventually become short circuits. Therefore there will be roll-off of the highest frequencies. If this roll-off happens too abruptly it will cause a massive phase shift and cause oscillation. The thing to do is to control that roll-off early before it gets that bad, and C5 (68pf) is chosen for that purpose. It provides increasing negative feedback with frequency around the entire circuit to make the gain bandwidth product 1.8Mhz. This gives rise to a high frequency response (-3dB point) at 1.8Mhz / 15 (the gain) which is 120kHz, and that's roughly what the simulation showed.

Rod Elliot's version using designated low noise transistors measured -127dB EIN (equivalent input noise) which using a -67dB MC cartridge (which many are) gives a signal to noise ratio of 60dB. As vinyl is around -56dB (quiet vinyl that is) then 60dB should be adequate.

The gain in this circuit is lower but the transistors are not designated low noise so on balance I should expect something similar.

I'm not sure about distortion though. Open loop a single transistor can be very distorted and I'm guessing around 10%. With 40dB negative feedback I'm hoping for 0.1% or better (10%/100 (the 40dB loop gain)).

If you have a couple of 9V batteries, the parts and some stripboard you may want to build the circuit (you'll need two for stereo) and beat me to it. Do let me know what you think.

One last point regarding my choice of capacitors for C1, 2 and 3. These block DC by the way. C1 is chosen for a low frequency roll off around 1Hz (-3dB) pushing the phase shift and any turnover distortion below our hearing capabilities. C2 in the bootstrap should result in one tenth of that frequency or it will help to cause a peak or blip in the bass response, which sounds dreadful. C3 needs to be chosen so that the 1/10th frequency is passed to the negative feedback or it will do the same. C3 with the expected load of 47k (labelled R8 here) and the overall negative feedback resistor R1 (18k) puts that frequency half to a third lower. As such the -3dB point is very close to 45 degrees which it should be.

Next is build and test results. I do hope the noise and distortion will be within acceptable limits. It may take me some days to get around to that.
Not simple enough for Google-Bot to understand...
Back to Top
 Post Reply Post Reply Page  <12345 10>
  Share Topic   

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.01
Copyright ©2001-2018 Web Wiz Ltd.

This page was generated in 0.078 seconds.