I stumbled onto Rod Elliot's site and found a very simple mic pre-amp that I'm really impressed with.
Check out: Low Noise Microphone Preamp
Two NPN trasistors, the first bootstrap coupled to second so that the second acts in a constant current mode. This results in very low noise. The output is fed back to the input reducing THD. At small signal levels (the important ones) sources like vocals for example, ~ 20mv, produce THD less than 0.01% in simulation. Even at higher levels close to the compliance the THD stays below 0.1%. Indeed simulation with TI's Tina shows it meets' Rod's claims.
I built one up using a pair of 2N5210s and it actually outperforms simulation in the dynamic range department.
As shown the input Z is 1.2k, and the gain is 40.
Ok, perhaps not quite enough gain for a ribbon mic but certainly enough for most commonly used modern mics.
It would require some gain reduction to amplify loud sources like close miced guitar amps or drums. Easy - reduce the value of the feedback resistor or increase the value of the input resistor.
It's single supply so uses three 10uf coupling caps. I see no problem using electrolytics. Perhaps the one cap up front could be a film type if you reduce from 10uf to 1 uf for cost and size reasons. This affects the low corner favorably in fact filtering out rumble.
Now should I use an input transformer?
The only real advantage I see is isolation from potential ground loops that might occur in a live big stage situation, otherwise one needs to spend quite a bit $ to get a decent wide band higher voltage tolerant transformer worthy of the performance of the circuit without.
I'm inclined not to use a input transformer since I will be using a four of these in my home recording gear where I can control loops etc.
Any opinions on that?
Check out: Low Noise Microphone Preamp
Two NPN trasistors, the first bootstrap coupled to second so that the second acts in a constant current mode. This results in very low noise. The output is fed back to the input reducing THD. At small signal levels (the important ones) sources like vocals for example, ~ 20mv, produce THD less than 0.01% in simulation. Even at higher levels close to the compliance the THD stays below 0.1%. Indeed simulation with TI's Tina shows it meets' Rod's claims.
I built one up using a pair of 2N5210s and it actually outperforms simulation in the dynamic range department.
As shown the input Z is 1.2k, and the gain is 40.
Ok, perhaps not quite enough gain for a ribbon mic but certainly enough for most commonly used modern mics.
It would require some gain reduction to amplify loud sources like close miced guitar amps or drums. Easy - reduce the value of the feedback resistor or increase the value of the input resistor.
It's single supply so uses three 10uf coupling caps. I see no problem using electrolytics. Perhaps the one cap up front could be a film type if you reduce from 10uf to 1 uf for cost and size reasons. This affects the low corner favorably in fact filtering out rumble.
Now should I use an input transformer?
The only real advantage I see is isolation from potential ground loops that might occur in a live big stage situation, otherwise one needs to spend quite a bit $ to get a decent wide band higher voltage tolerant transformer worthy of the performance of the circuit without.
I'm inclined not to use a input transformer since I will be using a four of these in my home recording gear where I can control loops etc.
Any opinions on that?
If the four mics dont touch each other, there will be no ground loops.
I have build a 8 channel version. nice and simple circuit indeed, but not low noise.
I have build a 8 channel version. nice and simple circuit indeed, but not low noise.
Have you determined the source of noise Tshrama?
Simulation shows noise down in the the 20uv range for the audio band. My equipment isn't good enough to measure noise down that low so I certainly don't see it on the bench. However, I am operating from a well filtered analog bench supply. Perhaps I'll discover different when I use a small transformer/bridge filter.
The circuit consumes ~3ma at 35 volts. It's quite easy to add series R and caps, etc pi filter at that low a current draw should the source of the noise be power supply. One could easily afford 30-50 ohms of series R and high cap filter values at that draw for example.
Simulation shows noise down in the the 20uv range for the audio band. My equipment isn't good enough to measure noise down that low so I certainly don't see it on the bench. However, I am operating from a well filtered analog bench supply. Perhaps I'll discover different when I use a small transformer/bridge filter.
The circuit consumes ~3ma at 35 volts. It's quite easy to add series R and caps, etc pi filter at that low a current draw should the source of the noise be power supply. One could easily afford 30-50 ohms of series R and high cap filter values at that draw for example.
Without a transformer it does not have differential (balanced) inputs. I consider this to be almost mandatory.
A less simple design (6 transistors) that I came up with that has very good performance:
Solid-State Balanced Microphone Preamplifier
Edit: A buildable schematic (and gerber files) is shown somewhere on page 6-9. Ignore the early schematics, they are untested.
The design can be configured for between about 2 and 70 dB of gain if so desired.
I found that with a Jensen output transformer it could match the performance (on the test bench at least) of my Focusrite ISA One preamp. The output transformer doesn't need to be an expensive one. Alternatively, you could use an op-amp in place of the output transformer.
The input transformer is completely optional in this design, whereas it is almost mandatory in Rod Elliot's design IMO.
There are lots of good and simple transformerless mic preamp designs. THAT company has some great references on their site about designing microphone preamps that I highly recommend reading.
A less simple design (6 transistors) that I came up with that has very good performance:
Solid-State Balanced Microphone Preamplifier
Edit: A buildable schematic (and gerber files) is shown somewhere on page 6-9. Ignore the early schematics, they are untested.
The design can be configured for between about 2 and 70 dB of gain if so desired.
I found that with a Jensen output transformer it could match the performance (on the test bench at least) of my Focusrite ISA One preamp. The output transformer doesn't need to be an expensive one. Alternatively, you could use an op-amp in place of the output transformer.
The input transformer is completely optional in this design, whereas it is almost mandatory in Rod Elliot's design IMO.
There are lots of good and simple transformerless mic preamp designs. THAT company has some great references on their site about designing microphone preamps that I highly recommend reading.
Here's the seminar on the THAT website that is worth reading.
http://www.thatcorp.com/datashts/AES129_Designing_Mic_Preamps.pdf
http://www.thatcorp.com/datashts/AES129_Designing_Mic_Preamps.pdf
Yeh I realize conventional wisdom dictates the use of a transformer with single ended input but I'll be damned if I can demonstrate the need either via simulation or with a built prototype. Both give excellent low noise results without.
This is the point of my post.
Anyone else care to simulate and come up with a different result?
BTW I also have a discrete op-amp transformer less/capacitor less, differential input design using seven transistors that has great performance with 65db of gain (or less by fb resistor selection) and can tolerate +/- 50v supplies but the part count is 5x of this simple design and uses a pretty expensive matched FET pair up front.
This is the point of my post.
Anyone else care to simulate and come up with a different result?
BTW I also have a discrete op-amp transformer less/capacitor less, differential input design using seven transistors that has great performance with 65db of gain (or less by fb resistor selection) and can tolerate +/- 50v supplies but the part count is 5x of this simple design and uses a pretty expensive matched FET pair up front.
Rod Eliots' work fine for recording of drums, electric guitar, singing etc. Especiallly with high sensitive, neodinium, dynamic mics.
Spoken word with ribbon gives too much noise. The BJT type isnt a low noise type, and the 1K2 input reaiator immidiately adds 4nV/sqrtHz.
I have build 2 transistor preamo, gain 40dB with l3ss than 0.5nV/sqrtHz.
For most dynamic mics, you can reduce the inout R from 1k2 to 300R. If you then chose a ZTX medium power BJT and you might achive 2nV/sqrtHz.
Spoken word with ribbon gives too much noise. The BJT type isnt a low noise type, and the 1K2 input reaiator immidiately adds 4nV/sqrtHz.
I have build 2 transistor preamo, gain 40dB with l3ss than 0.5nV/sqrtHz.
For most dynamic mics, you can reduce the inout R from 1k2 to 300R. If you then chose a ZTX medium power BJT and you might achive 2nV/sqrtHz.
Which is why just about nobody seriously considers inverting amplifier topologies where noise performance is a primary concern - like dynamic / ribbon mic or phono preamps.
Mind you, this one still beats your average Realtek onboard audio mic input by roughly an order of magnitude. It should still be on par with or slightly better than a non-inverting NE5532, too (and definitely better than a TL071 or 4558 based circuit). So if that's your kind of standard, then yes, it does actually qualify as "low-noise". Even the lowest-noise mic preamp in the world would only ever be 6 dB better with a 600 ohm dynamic mic, or 11 dB at 150 ohms.
The circuit from the Soundcraft 200 mixer (essentially a simpler version of ESP P66, see Soundcraft 1600 thread in Analog Line Level) will match the same level of noise performance if turned down to a gain of 31 dB (36).
Presumably non-inverting with a carefully-chosen input transistor then. I do think 40 dB is pushing it a bit with two transistors, a third may come in quite handy.
The downside of any unbalanced input mic preamp is that you probably don't want to use it with longer cable runs unless your mic cabling is coax, which in turn tends to be less flexible than you'd like... typical studio/pro mic cables are all twisted pair for use with balanced inputs. And while such an amp can be set up to supply an electret capsule, anything that requires phantom power is out.
Simulation reports the noise rising parabolic ally with frequency which is typical. It's about 25uv at 6khz. That's about where my 63 year old ears cut out - perfect.
I really like simplicity. Multiplying the number of components by four and the cost by the same factor just doesn't make sense to me since I won't hear it anyway.
My general approach is "The less number of PN junctions the better", and that's why I disqualify op-amp ICs.
I really like simplicity. Multiplying the number of components by four and the cost by the same factor just doesn't make sense to me since I won't hear it anyway.
My general approach is "The less number of PN junctions the better", and that's why I disqualify op-amp ICs.