Good Phono preamp circuit?

[nuvistor]

A thread about an MC phono pre with discrete input and op amp output, passive/active EQ is here, could be modified for MM. Input is single-ended so expect tweaking for DC operating point.

If you have NE5534s and matched discrete transistor pairs you can disable the on-chip bipolar input pair and connect a discrete differential pair as I described here, OL gain will be lower and stability compensation is probably easier than amplifierguru's example. I highly recommend cascoding the input pair to reduce common-mode effects.

If you're going to the trouble of adding discretes to an op amp you might also consider adding an output buffer using BUF634 or discretes, this uses discretes, you could class A bias the op amp output stage by removing Q1 and placing D1 in series with D2, and smaller output transistors like 2N4401/4403 would be OK.

[PeterMoreton]

It seems that the 2SK146 has been superceded by the Toshiba 2SK389, which is supposed to have a slightly better spec, and can be ordered easily (at least in the UK).

If I can actually place an order for either the 2SK389 or 2SK146, I'll redesign my circuit around these.

[pooge]

Although not a simple circuit, Robert Grodinsky patented a passive/active RIAA circuit back in 1982. It is Patent No. 4,312,060. Here is the abstract:

The preamplifier has a first stage or section of low or unity gain, such as an emitter-follower circuit, which also isolates the preamplifier from impedance variations of the cartridge, and with a bandwidth for greater than the audio range to reproduce signals with a very high amplitude, rapidly changing wavefront without slew or amplitude overload distortion. A RIAA high frequency de-emphasis network is coupled between this first stage or section and a second amplifier section. A RIAA low frequency emphasis network is preferably placed in a negative feedback circuit of the latter amplifier section where the feedback circuit does not contain the high frequency components filtered out by the high frequency de-emphasis network. The second amplifier section need not have the bandwidth of the first amplifier stage or section.


Borbebly adopted a similar split feedback topology in the late 80s. He just published a recent article on it, but I haven't read it yet. His new article claims a low cost version, but his idea of low cost and mine differ. I doubt it deals with the time constant mentioned above in the thread. Anyhow, his web site has the published articles:


Borbely articles

[amplifierguru]

Hi Peter,

The 2SK146 (dual N fet I suggested because mlloyd has some) can be substituted with 2SK389 or a pair of 2SK170. The design I gave does not need selected GR,BL, or V grading just a pair the same - the duals are generally within 10%. For the 2SK170's, if they're not a perfect match then 22ohm or smaller emitter resistors should balance them up. If you want offset adjustment the op amp has this facility.

Hi Nuvistor,

I note your comments re the 5534 and your link and suggestion to cascode - but I like to use all that I've paid for rather than disable one stage. The discrete diff'lly loaded stage feeding , say, a 5534 front end should be fine despite it's unknown CMRR - it will cope to the error tolerance of the 3K9 source resistors.

If you're wanting Class A output at 3mA see pic. Probably not so good as will limit negative swing unless made much higher Iq. especially when it comes to driving the hf low Z RIAA network - the 5534 would do better with just the pull down R to -Vs.

If you wanted all discrete the op amp could be replaced by a second diff'l stage with current mirror load then the Class A output.

lot's of fun without the noise.

Cheers,
Greg

[mlloyd1]

sorry,

these devices are pretty much unobtanium these days. Stick with 2SK389/2SK369. I just happen to have a few purchased a long time ago when i was single and could buy expensive semiconductors without somebody looking over my shoulder.

oops, oh hi honey ...


and thanks ampguru. i'll give it a shot.

mlloyd1

Originally posted by amplifierguru
...
The 2SK146 (dual N fet I suggested because mlloyd has some) can be substituted with 2SK389 or a pair of 2SK170. ..

lot's of fun without the noise.

Cheers,
Greg [/B]

[nuvistor]

I note your comments re the 5534 and your link and suggestion to cascode - but I like to use all that I've paid for rather than disable one stage.

Even with the added burden of the extra lag and the (perhaps incidental and unneeded) opportunity for increased slew rate when the 5534 1st stage is off? With a discrete 1st stage followed by 5534 at low closed-loop gain I'd rather bypass the 5534 1st stage and not have to work so hard at stability compensation, after all single-stage active EQ RIAA forces a low closed loop gain above 20kHz, about 5 from 200 - 500 kHz to open loop gain intercept even with the 50kHz zero in the modified RIAA EQ.

With a 100MHz GBW op amp like LM6171 or AD817 in place of 5534 the lag below 10MHz is much less and you don't have access to the 1st stage collector nodes so disabling 1st stage is not an option, higher open-loop gain should be possible, for compensation I would consider NFB around the op amp for Miller-integrator compensation with some source degeneration if noise requirements permit.

If you're wanting Class A output at 3mA see pic.

5534 uses NPNs for output source/sink with feedback around the lower NPN, I've driven discrete output stages from 5534 comp pin 5, 5mA into this pin from V+ biases 5534's lower NPN on and also can bias a Vbe multiplier or diode string for class AB, with 2N4401/4403 you have a pretty good headphone driver, or just use pin 5 as output if no short circuit protection is OK. 5534 pin 5 is at lower NPN collector, sinks to negative rail, output pin 6 adds a diode and resistor. Sure it's a small step to a completely discrete op amp, but with IC op amps there's less to build and 5534 smoke is cheap. If 500uA 1st stage tail current is enough, 1st stage load resistors can be the 5534's internal 13k to V+. For inverting gain common-mode is not an issue and a simple resistive tail is OK, for non-inverting I would use an active current sink.

[amplifierguru]

Hi Nuvistor,

You know your 5534's - I thought using these was a good idea as they are cheap as chips, universal and good output drive.

The RC comp is simple - starts at 10KHz and back to unity by 1.5 MHz but gives an extra 40 odd dB at audio frequencies - in partic at LF where , with high gain MC gain at 80dB, it's needed for NFB. At the other end the MM gain of say 40dB means 12 dB or (x4) from 50KHz up - so the decomp 5534 (no 22pF) will be stable to 7 MHz.


cheers,
Greg

[nuvistor]

Hi ampguru -

Ye olde op amp chippes are good for experiments where damage risk is high.

I agree with your numbers as given, and that you would need the extra OL gain for MC. First stage slew rate is 4mA/2nF or 2V/us, should be OK for phono stage. 2n2 compensation could be reduced for MC.

[zinsula]

Take a look:
http://www.users.nac.net/markowitzgd/phonopre.html

There are some threads about this phono pre here, eg this:
ultrasimple mm/mc riaa preamp

When you need MC pre, do not use 2SK170 instead of the 2SK147 (out of production, look at the prices from Borbely... ) on that schematic, but use 2SK369 (which has same specs as 2SK147 except power handling). It has lower noise at low source resistance (=MC) than 2SK170. See attached graphs. And it is cheap.

Btw, same goes for 2SK146 vs 2SK389, they are not really the same. It's only that both are out of production now....

Tino

[mlloyd1]

Well, maybe I exaggerated a bit. I don't want to unnecessarily scare off anybody that may be determined to try some of the ideas here.

While it is true that Toshiba no longer makes the 2SK146, 2SK147 devices and others have pointed out some (expensive) sources for NOS....
There is a thread on this board somewhere that mentions new product of similar (or better) specs from either InterFET or Linear Systems, I can't remember which one. You can get duals or singles. They're probably not cheap either, but at least they are new production devices.

However, no P channel complements, unfortunately.

mlloyd1
================
[edit] ok, I found them:
http://www.interfet.com/pdf/DS_IFN146.pdf
http://www.interfet.com/pdf/DS_IFN147.pdf

group buy anyone?
================

Originally posted by zinsula
...