Long time builder, inexperienced designer. Attempting to make a simple low noise mic pre, with a high impedance input suited for dynamics & ribbons. Came across this circuit by Bob Cordell, dual jfet LSK489 with presumably an impedance in the many many mega-ohms. Anyone built one of these or something similar?
https://www.cordellaudio.com/JFETs/LSK489appnote.pdf page 11/12
I’m mainly trying to figure out the line driver, be nice to use both halves of the 4562 or I could just whack a THAT1606 on the end. Or use a 1646 instead of the 4562 entirely but it might be too hungry coming straight off the transistors without an opamp buffer? Or of course simply do an impedance matched driver after the existing buffer like so
The other thing is this pre apparently provides 60dB, it’d be nice to bring that up to 70. I’m not sure if that’s better squeezed out of the transistors at the front end or done at the end with an opamp?
Any thoughts? Thanks in advance
https://www.cordellaudio.com/JFETs/LSK489appnote.pdf page 11/12
I’m mainly trying to figure out the line driver, be nice to use both halves of the 4562 or I could just whack a THAT1606 on the end. Or use a 1646 instead of the 4562 entirely but it might be too hungry coming straight off the transistors without an opamp buffer? Or of course simply do an impedance matched driver after the existing buffer like so
The other thing is this pre apparently provides 60dB, it’d be nice to bring that up to 70. I’m not sure if that’s better squeezed out of the transistors at the front end or done at the end with an opamp?
Any thoughts? Thanks in advance
Hello Tashi1, I designed an instrument preamp design using the LSK489 originally, but shifted over to the LSK389. have a look at the updated data sheets for both, I think the noise performance was/is very good with the LSK389 now. IMHO, get the extra 10 dB after the FET input stage, once you are safely out of the noise floor, with a low noise op amp. Cheers!
Just some suggestions, but if you are looking to design a mic preamp that will work with a variety of microphone types, including condenser mics, you could/should study some of the preamp designs of Neve, Neumann, Bruel and Kjaer etc.
The Neve 1073 is a classic, there are plenty of kits and parts out there. You may want to use an input transformer and incorporate a 48V phantom supply (switchable). Your balanced line out could also incorporate a signal transformer for isolation. Perhaps a stepped attenuator within the preamp section, and maybe a send /return for a EQ section. A lot of preamps also have a Line/Instrument level input which basically inserts after the first stage amplification. Maybe it's too much for what you want to build right now, but in the future perhaps you may want to add in some features. Cheers!
The Neve 1073 is a classic, there are plenty of kits and parts out there. You may want to use an input transformer and incorporate a 48V phantom supply (switchable). Your balanced line out could also incorporate a signal transformer for isolation. Perhaps a stepped attenuator within the preamp section, and maybe a send /return for a EQ section. A lot of preamps also have a Line/Instrument level input which basically inserts after the first stage amplification. Maybe it's too much for what you want to build right now, but in the future perhaps you may want to add in some features. Cheers!
Thanks, good idea. I was hoping to use this as an instrument amp also! DIs seem a bit like an extra step, if I can get a high enough impedance preamp or line amp, I can skip them entirely. Care to share a schematic?
That’s exactly why, because those mics have low impedance (low current available). Have you tried increasing the load on an old hand crank generator and it gets harder to turn? The same thing happens to a ribbon or moving coil diaphragm, killing the high frequencies first.
You want a high impedance preamp that won’t draw much current, so you won’t load the mic. Hence some of the best regarded ribbon pres are the likes of Gordon 5 & AEA TRP, very high impedance inputs.
So I went down the rabbit hole, seems like the best sounding FET designs are based on Graeme Cohen’s double balanced topology. Found Fred Forssell’s take on it here:
https://web.archive.org/web/20040405175025if_/http://www.forsselltech.com:80/JFETMP1.PDF
and the excellent performing implementation by Samuel Groner as the Monte Generosi:
http://www.nanovolt.ch/resources/microphone_preamplifiers/pdf/monte_generoso_r1.pdf
Also turns out the on chip preamps have pretty solid performance too, so to keep things very simple I’ll start with something based on the INA163. 2M input impedance and very low noise. I’ll draw it up and post the schematic to see if I’m on the right track.
I’ll make a Cohen design for the next project!
https://web.archive.org/web/20040405175025if_/http://www.forsselltech.com:80/JFETMP1.PDF
and the excellent performing implementation by Samuel Groner as the Monte Generosi:
http://www.nanovolt.ch/resources/microphone_preamplifiers/pdf/monte_generoso_r1.pdf
Also turns out the on chip preamps have pretty solid performance too, so to keep things very simple I’ll start with something based on the INA163. 2M input impedance and very low noise. I’ll draw it up and post the schematic to see if I’m on the right track.
I’ll make a Cohen design for the next project!
Looking at the circuits that were posted previously I would also suggest the THAT1580. I have only made very good experiences with it and it has very low noise.
You are free to use it with a pot or you also have the option to pair it with a THAT5173 and you can control the gain through SPI,
You are free to use it with a pot or you also have the option to pair it with a THAT5173 and you can control the gain through SPI,
Hi!
In my experience with buiding a low noise pre-amp for a ribbon mic, as we decrease input impedance, high frequencies indeed are killed first.
But, above 10kOhm input impedance, it makes not much difference. Below 4.7kOhms we clearly can see the difference showing up.
See attached some measurements I've done some years ago comparing different input impedance values versus the baseline of 10Mohm input impedance.
10k, 4,7k, 2.2k, 1k and 560Ohms. Charts zooming 10 to 10kHz, 100 to 10kHz, 1k to 10kHz and 4kHz to 10kHz so as to better observe the frequency response in each range.
These measurements must be observed from the relative perspective: same mic, same pre-amp circuit and same loudspeaker. The only variable changed was the insertion of a resistor at the input so as to vary the input impedance.
I ended up building this pre-amp with 2 pairs of JFET LSK389 (parallel to reduce noise) with 100kOhm input impedance to gnd.
Hi Tashi1, Hope you are not offended but I would prefer not to share the final schematic of my design.. as I now started to build these and offer them as a "commercial" product.
But the preamp is performing in Class A , push-pull and drives a small signal transformer at its output. The LSK series is well suited because of the matching of the pair of FETs, and they are indeed very quiet. I went for about 24 dB of gain (variable feedback) and at full gain the noise floor is super low, I think better than 110 dB signal to noise ratio. For instrument levels I felt that 0 to 24dB is more than enough range for the gain. The transformer is not a super high end, though good enough for guitar and bass etc for the frequency response.
I have also tested it with higher end output transformer with very good results. For your mic /instrument preamp, you'd probably want to use something like the Jensen, Cine-Mag or Hammond studio series transfos. They are pricey but you'll get full audio bandwidth , balanced and isolated signal for DI applications.
If I were doing a preamp that would work for both mic and instrument, I would probably try to optimize the mic preamp front end separately from the instrument front end and merge the two at the middle or output section. You definitely don't "need" 60 dB of gain for the instrument level.
Cheers!
I didn't realize ribbon microphones normally have built-in step-up transformers. If the ribbon were driving the microphone preamplifier directly, the source impedance would be very low indeed and anything would do for the microphone preamplifier input impedance.
What does the impedance graph of a typical ribbon microphone (with built-in transformer) look like? It determines what kind of input stage would lead to least noise.
What does the impedance graph of a typical ribbon microphone (with built-in transformer) look like? It determines what kind of input stage would lead to least noise.
Hi! Yes, the ribbon in general has a stepup transformer.
See the one I used as a reference and see if my conclusions make sense.
The secondary winding DC resistance is around 4ohms - I measured at the mic final output leads.
I don't have the impedance curve - I'd have to think about how to measure it without damaging the mic.
But from load curves I took, this is what I observerd:
With 10kohms load, there is no change in the signal from 20 to 10kHz compared with 10Mohm load.
With 560ohms load, I see around 5 to 6dB drop starting around 200Hz-300Hz and up and it stays flat up to 10kHz.
It would indicate that impedance starts at very low levels (10ths of ohms), raises to around 500Ohms and stays there up to the high frequencies.
Zooming the high frequency range - we clearly see 5 to 6dB difference flat.
6dB in voltage is half voltage so the output impedance would be around the same of the input load applied (560Ohms)
See the one I used as a reference and see if my conclusions make sense.
The secondary winding DC resistance is around 4ohms - I measured at the mic final output leads.
I don't have the impedance curve - I'd have to think about how to measure it without damaging the mic.
But from load curves I took, this is what I observerd:
With 10kohms load, there is no change in the signal from 20 to 10kHz compared with 10Mohm load.
With 560ohms load, I see around 5 to 6dB drop starting around 200Hz-300Hz and up and it stays flat up to 10kHz.
It would indicate that impedance starts at very low levels (10ths of ohms), raises to around 500Ohms and stays there up to the high frequencies.
Zooming the high frequency range - we clearly see 5 to 6dB difference flat.
6dB in voltage is half voltage so the output impedance would be around the same of the input load applied (560Ohms)
Last edited:
That's accurate enough, I just wanted to know the order of magnitude. Measuring signal transformer DC resistance is usually not recommended because of possible magnitization of the core, by the way.
For a decent noise performance with a 500 ohm source, you can either use a bipolar transistor biased at about half a milliampere to a milliampere of collector current, or a high-transconductance (>> 2 mS) JFET.
As you probably shouldn't want to let the base current flow through the microphone because of possible transformer core magnetization, the bipolar transistor requires AC coupling. With bootstrap techniques and series feedback, you can still make the input impedance quite high, but using a JFET may be simpler.
The bipolar stage would have very poor noise performance with the high impedance of an electric guitar as source (too high equivalent input noise current due to base current shot noise). Bootstrapping doesn't help against that. A JFET would work fine with electric guitar impedance levels.
As you probably shouldn't want to let the base current flow through the microphone because of possible transformer core magnetization, the bipolar transistor requires AC coupling. With bootstrap techniques and series feedback, you can still make the input impedance quite high, but using a JFET may be simpler.
The bipolar stage would have very poor noise performance with the high impedance of an electric guitar as source (too high equivalent input noise current due to base current shot noise). Bootstrapping doesn't help against that. A JFET would work fine with electric guitar impedance levels.
The original post showing Bob Cordell's proposed design seems pretty reasonable to go with. It's not exactly a complete circuit, as it is showing the LSK 389/489 as an open input +/- differential pair. It is also leaving the current sinking parts as a black box..so you need to design that aspect.
A similar mic preamp design by Elliot Sound Products (Project 66) for low noise balanced microphone preamp, but uses bipolar transistors in the front end differential amplifier. I'd suggest having a look at that one to get some direction on tying in the differential pair +/- inputs to the microphone...for input impedance and grounding. Also the power supply to the first and second amplification stage should be very well regulated.
ESP Project 66 low noise mic preamp
Cheers
A similar mic preamp design by Elliot Sound Products (Project 66) for low noise balanced microphone preamp, but uses bipolar transistors in the front end differential amplifier. I'd suggest having a look at that one to get some direction on tying in the differential pair +/- inputs to the microphone...for input impedance and grounding. Also the power supply to the first and second amplification stage should be very well regulated.
ESP Project 66 low noise mic preamp
Cheers
One thing I havent seen mentioned is the ability to easily swap phase 0 to 180. Many acoustic guitar preamps have this feature, as sometimes you get into an acoustic "situation" and that cures it. Unless of course this preamp will only be used for making recordings...
For the Cordell circuit, I was thinking that the the two current sinks would probably be good to work with a current mirror so one leg (I2) is tracking the other (I1) current. I haven't set this circuit up on a breadboard to try it out though.. but I imagine you would want both current sinks to be balanced and stable with temperature.
Yesterday, I simulated the design in LTspice and I wasn't able to remove an oscilation that seems to have its origin in the input stage.