A Flat Moving Coil Preamp Using Paralleled Amplifiers

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by mediatechnology »

I know that now that I've done a flat moving magnet phono preamp someone will eventually ask for a moving coil design. This design presented below is a proof-of-concept that I recently had a chance to test.
It's inspired by our new forum member Scott Wurcer and his 1982 article "Stacked Amplifiers Lower Noise," Scott Wurcer, Charles Kitchin, EDN, October 13, 1982. (See: viewtopic.php?f=12&t=700)
Additional inspiration comes from Graeme Cohen's cross-coupled "double-balanced" cricuit of the same era.
I also want to thank Gary Hebert of THAT for his input.


Obtaining Low Rbe Transistors

One of the problems designing any active moving coil input are obtaining low Rbe transistors.
If we could get 2SB737s from a current manufacturer we'd be in good shape. But we can't.
Self and many others have had good results paralleling the far more common 2N4403 with its Rbe of 20 Ohms.
The BOM of Self's Figure 8.5 (not shown), requiring 3X 2N4403 and a dual op amp, is fairly low cost despite the large number of bits in the circuit.
Though there are other low Rbe transistors out there - some are equally hard to find - it generally takes more than one and a lot of support "stuff."
A fully-balanced implementation could easily double the number of parts.
In order to keep a MC input simple, low parts count, with reasonably good performance but not necessarily the lowest cost I ask: "What can be done with simple mic preamp ICs?"

Most of the moving coil preamp schematics that you see published are 20-35 dB gain "head amps" to be used prior to a moving magnet preamp input and RIAA stage.
A flat moving coil preamp driving a sound card for RAW recording needs a lot of un-equalized gain.
The gain estimate to drive a sound card input with no playback EQ is 45 to 60 dB, maybe more, maybe less.
Douglas Self's 2009 survey ofabout 85 moving coil carts found three clusters of typical MC levels centered around 100 uV, 200 uV and 500 uV at 5 cm/s.


What limits using mic preamp ICs?

Noise. Mic preamp ICs such as the SSM2019, THAT1510/1512 and INA217 don't have the noise performance for a moving coil phono preamp despite the datasheet bullet-points suggesting that they do.
These parts are optimized for higher source impedances in the 100-1000 Ohm range.
Stand-alone mic preamp ICs are still superior to the 5532-based MC head-amps that you do sometimes see in low-cost switchable designs.

An MC preamp needs low noise with 10-30 Ohm source impedances.
Some extreme MC carts are even lower.
Stacked, paralleled mic preamps might be able to get the noise down.

When paralleled, two ICs provide 3 dB noise reduction because the "input transistors" (actually the entire instrumentation amp) are paralleled.
Four devices, whether in parallel or cross-coupled, should give us a theoretical 6 dB noise improvement.
More than four parts and things get silly.

For a 6 dB increase in cost we get 3 dB of noise improvement.
For a 12 dB increase in cost we get 6 dB of noise improvement.

Is it worth it?
I suppose it depends.
Some carts may only need two. Others may require four.
The eight THAT1512 required to build a stereo unit are definitely cheaper than a MC step up transformer and MM preamp.

Can these easy-to-use mic preamps be stacked to lower noise to acceptable levels?
It looks like they can.

Image
A Flat High-Gain Moving Coil Phono Preamplifier Proof-of-Concept Using Multiple Stacked Amplifiers.

The preamp above has sufficient gain to bring moving coil levels up to "line" level for RAW, no-RIAA recording.
Using THAT 1512s maximum gain (unbalanced out) is 60 dB. Balanced gain is up to 66 dB.

The input is differential balanced and can also be used with unbalanced RCA terminations.
The output is balanced or single-ended depending on jumpers J1 and J2.
The drawing shows the input jumpers in a cross-coupled configuration to provide a balanced output.
J2, when strapped for unbalanced output, parallels all four 1512 outputs to drive the single-ended loads.

Measured Noise Performance

I tested the design in the balanced in/unbalanced out configuration due to unbalanced instrumentation. (No AP or D-Scope.)

With 60 dB of gain I measured output noise of -65.5 dBu for a single amp. (Each THAT1512 is within 0.5 dB.)
The combined stack of four THAT1512 is reading -70.5 dBu.
The noise improvement is a tad under 5 dB.

I'm using an HP400FL with the 100 kHz 6dB/octave filter engaged.
The noise measurement bandwidth is 157 kHz.

The total input noise works out to be -130.5 dBu (231 nV) or around 0.58 nV/√Hz with a 10 Ohm source impedance.
Scaling for a 20 kHz measurement bandwidth I calculate an input noise around 82 nV.
Not bad for an open, unshielded Protoboard.

A 100 uV output moving coil cart (5cm/s, 1kHz) should have an unweighted, un-EQ'd SN of about 62 dB.

I don't have a moving coil cartridge to test this but expect to score a Denon DL-103 soon.
The DL-103 has a 250 uV output, 14 Ohm impedance and is designed for 100R loads.

Circuit Description

The cartridge is direct-coupled to the input and when connected balanced floats from ground.
The input offset current for a single THAT1512 is +/-1.4 uA max.
The input offset current for multiple THAT1512s is likely to be the rms power sum.
I tried to measure the "cartridge" current.
The lowest voltage measurement in the last digit of my DVM is 10 uV.
It read 0 with an occasional last digit indicating it might be somewhere around or less than 10 uV.
Assuming that it was 10 uV, the bias current is on the order of a uA.
Suffice it to say that the combined offset, and the resulting cartridge current, is significantly less than the leakage current of a large-value electrolytic.

The input is fully-balanced with 100 Ohm differential termination. (Can be raised or lowered.)
Two series-connected 49R9 resistors provide cartridge loading and a bias path for the THAT1512.
An optional shared 1K "T-bias" resistor, Rcm, can be used to raise the common mode impedance and provides a return path for the sum of all devices' input bias and bias noise currents.
The common mode potential developed across the 1K Rcm in the prototype was 17 mV indicating that the sum of bias current for all eight inputs was about 17 uA.

Common mode rejection is provided internally within the THA1512s.
Because the majority of bias current noise and DC bias current is shared in Rcm and appears in common mode, it's noise contribution and DC offset are rejected.
Rcm has virtually no effect on noise and can be made higher in value limited by the DC CM offset it creates.
I've also tried 10K - but not with an actual cart.
(We've done tests up to 100K Rcm in 1510 mic preamps without seeing noise increases.)
Rcm can be jumpered out.
The benefit of either low or high common mode impedance when connected to a moving coil cart is not known at this time.

J1 and J2 determines the output configuration.
The schematic is shown jumpered for a balanced output with the input halves cross-coupled to produce anti-polarity output.
When a balanced output is used there is an additional 6 dB gain available.
For an unbalanced configuration J1 is jumpered to provide parallel drive along with J2 linked to parallel all four outputs.

The resistor in series with J4 serves as a "base stopper" for the 1512 when it is grounded for an unbalanced input.
I put it there as a precaution and its value is TBD probably around 5R.
Under some circumstances I've found the THAT1510 and INA217 will oscillate if one input is directly connected to ground.

Rg cannot be shared between devices.
Changing gain unfortunately requires changing the value of Rg in four places. (Eight for stereo!)
One could load the PC board with a single 1512, find the appropriate gain, and then stuff the remaining Rg resistors.
The preamp gain should be set to get it in the ballpark for the particular cartridge - a gain trim downstream will permit fine adjustment of level and channel balance.
I'm also considering jumpers (one for each 1512) to provide three different gains.

The output ballasting resistors, shown as 100 Ohm, are made relatively large for two reasons.
As shown Ro is 50 Ohms when balanced, 25 Ohms unbalanced.
(For Unbalanced outputs Rballast could be made 200R for an Ro of 50 Ohms.)

Reason #1 for the build-outs is to provide capacitive load isolation.
Reason #2 is DC offset between 1512s.

A larger build-out reduces the circulating current between outputs due to THAT1512 output offset.

At 60 dB gain, the maximum offset for a 1512 is 255 mV. (0.25mV x Gain + 5mV.)
Two 1512's with opposite offsets at the 255 mV limit may have up to 510 mV between them.
With 200R between outputs the standing output current is limited to 2.5 mA.
I suppose that higher offset devices are better because they produce outputs running in Class-A...

The output from the MC preamp will be AC-coupled downstream by a film cap in the EQ/monitor switcher.
If I were to do the flat MC preamp on a PC board I probably would make space for some bipolar electrolytics in the output for the person using only the preamp board.
In the prototype, with four random 1512s chosen, the net output offset with 60dB gain turned out to be about 35 mV.
No need for servos here.

Unless there's an error in my noise calculations, the 0.58nV/√Hz voltage noise density should get me in the ballpark for a pretty decent NF.
Though it uses a lot of THAT1512s, four per channel, they are readily available and little outboard circuitry is required.

This article shows a 1983-era balanced input phono preamp made from an instrumentation amp:
"Instrumentation Amplifiers Solve Unusual Design Problems," Scott Wurcer and Walt Jung, Analog Devices AN-245. http://www.analog.com/media/en/technica ... 1AN245.pdf
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by ricardo »

mediatechnology wrote:Douglas Self's 2009 survey ofabout 85 moving coil carts found three clusters of typical MC levels centered around 100 uV, 200 uV and 500 uV at 5 cm/s.
Where did he say this? Is it a public link?

Any idea of the Rs of these 3 groups of MC cartridges?
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by mediatechnology »

Image
"The output levels for 85 moving-coil cartridges at 5 cm/s," Douglas Self, Small Signal Design. http://www.amazon.com/Small-Signal-Audi ... 415709733/

The highest clusters are in the 150-199 uV, 250 uV and 500 uV levels.
For the low value I chose 100 uV rather than 150uV and 200 uV rather than 250.
That makes the steps a more even 1, 2, 5 sequence or +60dB, +54 and +46.
Of course it can be anything with an Rg equal to or greater than 5 Ohms.

Self's survey found moving-coil cart source impedance ranging from 1 Ohm to 160 Ohm with loads of 3 Ohms to 47K.
Self's conclusion was 100 Ohm loads will work with most, 500 Ohms almost all carts except the very high output ones requiring 47K. (Denon DL-110 and DL-160.)
Gold
Posts: 678
Joined: Sat Dec 01, 2007 5:20 pm

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by Gold »

ricardo wrote: Any idea of the Rs of these 3 groups of MC cartridges?
Off the top of my head I'd say that an Ortofon SPU would be in the low output category with a very low source impedance. A Denon 103 would be in the medium category and the DL-110 that Wayne mentioned would be in the high category. The SPU is my favorite.

I'm using a Denon 103. I have a 1:16 step up transformer loaded with a 3.5k ? resistor. I'll check the value when I'm in the studio. I would have gone higher in value (because playback is still a little bright) but the Shure SE22 ran out of gain. I'll look at the manual to see how much gain it has. I think around 70dB. It's a standard MM pre with a 47k termination.

I almost forgot. Fantastic Wayne, as usual. It might work well as a tape head preamp. I'll look at the impedance of the Studer heads I have.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by JR. »

Gold wrote:
ricardo wrote: Any idea of the Rs of these 3 groups of MC cartridges?
Off the top of my head I'd say that an Ortofon SPU would be in the low output category with a very low source impedance. A Denon 103 would be in the medium category and the DL-110 that Wayne mentioned would be in the high category. The SPU is my favorite.

I'm using a Denon 103. I have a 1:16 step up transformer loaded with a 3.5k ? resistor. I'll check the value when I'm in the studio. I would have gone higher in value (because playback is still a little bright) but the Shure SE22 ran out of gain. I'll look at the manual to see how much gain it has. I think around 70dB. It's a standard MM pre with a 47k termination.

I almost forgot. Fantastic Wayne, as usual. It might work well as a tape head preamp. I'll look at the impedance of the Studer heads I have.
I am generally not a fan of audio transformers, and even made a solid state MC version of one of my preamp kits back in the day. That said an audio step up transformer could really help dealing with such low Z transducers. Since we are dealing with such low power levels the transformer doesn't need to be huge.. A 1:16x transformer is shifting the impedance 256x 8-)

Looking at that list of MC cart manufacturers, how many are still making them today?

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by mediatechnology »

Looking at that list of MC cart manufacturers, how many are still making them today?
Apparently lots of people: http://www.amazon.com/s/ref=sr_pg_1?rh= ... B0007DA21W

Some are insanely expensive but the Denon DL-103 (not the newer DL-103R) is pretty reasonable comparing in price to a good MM.

My 4X THAT1512 MC preamp has a noise penalty built in due to being fully-balanced but I have to think - as Paul has pointed out in the MM thread - that the added hum and buzz rejection outweigh the slight noise disadvantage.
To realize the benefit a fully-balanced tonearm cable is required but that isn't hard to do if the arm has a DIN connector.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by JR. »

Looking at the prices for some of those MC makes my old $99 kit price seem rudely inappropriate.

=====

For balanced vs SE noise, yes ein voltage increases (incoherently IIRC), but the analysis of input noise current is a little different, especially if the input source impedance shunting between the two inputs is low, relative to the impedance to ground from each input. Note; the termination resistance can be in shunt too so ratio of shunt impedance vs impedance to ground can be large. .

For noise analysis noise caused by the noise current present at each input is also getting fed into the opposite input with only a little attenuation caused by the voltage divider, and will mostly cancel out.

I may not be describing this clearly but for a low impedance source, with significant input noise current, its not as bad as first appears due to some cross cancellation.

JR
Cancel the "cancel culture", do not support mob hatred.
Gold
Posts: 678
Joined: Sat Dec 01, 2007 5:20 pm

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by Gold »

JR. wrote:
Gold wrote:
ricardo wrote: I am generally not a fan of audio transformers, and even made a solid state MC version of one of my preamp kits back in the day. That said an audio step up transformer could really help dealing with such low Z transducers. Since we are dealing with such low power levels the transformer doesn't need to be huge.. A 1:16x transformer is shifting the impedance 256x 8-)
A lundahl transformer is no compromise. A 1;1 Lundahl is essentially transparent at any reasonable signal level. When Wayne has this done I'll get both head amps and compare the dedicated MC to the MM with a step up transformer.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by mediatechnology »

For noise analysis noise caused by the noise current present at each input is also getting fed into the opposite input with only a little attenuation caused by the voltage divider, and will mostly cancel out.
I do get what you're saying.
Yes, that's how the T-bias magic works (for current noise) when Rcm >> than Rdiff. (See: viewtopic.php?f=6&t=191)
I sent Rosalfonso Bortoni a T-bias preamp based on a 1510 and he ran it on the AP with Rcm ranging from 0R to 100K.
I don't recall the Rbias but it was likely 1-1K5.
There was no discernible difference in noise with the CM stage internal to the 1510 rejecting the noise having a typical CMRR of 50 dB.
The limiting upper bound for Rcm is the DC bias current that develops in common mode reduces input headroom.
At 34 uA maximum for both inputs a 100K Rcm has 3.4 VDC Vcm.
The reason for having T-bias in an AC-coupled design are reduced input capacitor matching for increased LF CMR.

MC carts may benefit too.
I happen to think that with a real MC cart there may be some advantage to the higher CM impedance.
It's more "microphone-like" than a floating MM cartridge.
It also seems that interconnect source impedance imbalance at the very low impedance levels would benefit from higher Rcm.

The MM carts seem to prefer a low CM impedance when balanced but I want to re-check it now that it's a real PC board.
For balanced vs SE noise, yes ein voltage increases (incoherently IIRC)
Self, for single ended differential amp MC topology puts the voltage noise penalty at 2.8 dB compared to an single-ended stage.
I'm not sure why it's 2.8 dB and not three.

EDIT: Just read Paul's post.

Here's the data on the Lundahl LL1941 (1+1):(16+16) : http://www.lundahl.se/wp-content/upload ... s/1941.pdf
And the LL1943: http://www.lundahl.se/wp-content/upload ... s/1943.pdf
Both are available in silver.
All Lundahl pdfs: http://www.lundahl.se/datasheets/
Love Lundahls' Wallpaper.
Gold
Posts: 678
Joined: Sat Dec 01, 2007 5:20 pm

Re: A Flat Moving Coil Preamp Using Paralleled Amplifiers

Post by Gold »

I'm using the Lundahl 1678. 1:16 for the Denon 103. I have a 3k83Ω termination resistor in parallel with the Shure SE22 47kΩ input termination resistor. The SE22 data sheet says there is 54dB of gain. I'm using at least 50dB gain in this setup to get 7cm/sec lateral to +4dBu.

I have an extra set of 1678's currently wired 1:32 for the Ortofon SPU. I'd be happy to send them to you to check out if you think it would be useful.
Post Reply