Show posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Show posts Menu

Topics - Loudthud

Preamps and Effects / The JFET Bender Preamp
April 17, 2015, 05:54:14 PM
This thread is not about a distortion pedal called the Tone Bender.
It is about a preamp that emulates a Blackface Fender tube preamp.

In a previous thread I explored ways of getting opamps to act more like a 12AX7. Link: This is a continuation of the same idea except a JFET is used in the feedback network. A JFET is ideal for this circuit because it doesn't have a dead zone where it stops conducting current if the bias is set right. I hope some of you will try this idea in some exsisting designs to add some low level even order distortion. I call it the JFET Bender because it bends the straight line input-output relationship an opamp normally produces.

The JFET is used as a variable resistor in a region where it's resistance changes at different current levels. MPF102 JFETs were used because they had a bendable resistance in the right range. I originally tried 2N4391-3 but the pot setting was way too touchy. I had to install multi turn pots. The MPF102s have an Idss usually between 8 and 12mA. The Curve Tracer shot shows how Voltage and Current behave around zero volts. When the Voltage swings negative, the curve is fairly linear but as it swings positive, the JFET starts to go into constant current mode and the current levels out. Adjusting the Gate Voltage lets you adjust the amount of bend and thus the amount of distortion.

The distortion generated is mostly pure 2nd order at low output levels. Some 3rd order creeps in as the output of the stage reaches 10V pk-pk. The distortion rises to about 10% (this is adjustable via the pot) when the output of the opamp starts to clip. Because the gain is not the same on plus and minus halves of the waveform, one side clips before the other. To adjust the pot, set the pot near zero Volts. That's where maximum gain is. Apply a signal and adjust the pot until the gain is reduced to about 80% (-2dB) the max value. Or just adjust it by ear for a little gain reduction. If the pot is set too far negative, the JFET cuts off and the gain falls to 1. You can find a place where there is severe clipping on one side of the waveform if desired.

For those paying attention, the gain reduction caused by an N channel JFET in the first experiment is backwards compared to what it's like in the first stage of a tube guitar amp. What is really needed is max gain on positive inputs (closer to vertical line on the curve tracer) and a slow reduction in gain for negative inputs (the curve of the JFET lays over horizontally). I played guitar through a couple of stages setup backwards like this with a tone stack in between and it didn't sound right. It just didn't have the Fendery clean tone I was looking for.

Next I tried a J201 connected differently to invert the distortion. The Drain is grounded. When the Gate and Source swing positive, you have the bulk resistance Rds(on) in parallel with the Gate to channel diode. This is fairly linear until the diode starts to conduct. When the Gate and Source are pulled negative, you get the Idss curve and it lays over (resistance increases) in the third quadrant. The disadvantage of this circuit is that there is no adjustment and the resistors in the feedback network have to be adjusted to get the current and voltage in the right range for the J201. This circuit sounded better as a first stage but there was not enough distortion at the gain I wanted for the first stage. Distortion was in the range of 1% to 4% at 1Vrms input depending on the impedance in the feedback network. Too big a resistance in series with the JFET in the shunt arm of the feedback network limits the amount of distortion the JFET can generate.

Lastly I looked for a P-channel JFET that had favorable characteristics. J174 and J175 were nowhere close. The 2N5462 looked good. It is a near compliment to the MPF102. This worked in the original circuit, all that was needed was to tweek the pot for positive bias on the Gate. A 2N4343 was also used, it worked about the same.

Since a tube preamp has mostly inverting stages, a non-inverting opamp preamp will have to flip the distortion on alternate stages. I think not doing this is why the first try didn't sound very good. So a P-channel JFET is in the first stage and an N-channel JFET is used in the second stage. The second stage needs a gain around 50.

When the gain is high, there is not much Voltage available for the JFET to distort. On the second stage the feedback resistors set the gain at 100 and the pot is adjusted to get 6dB of gain reduction. This gets the distortion up to around 10% before the opamp starts to clip.

The preamp is configured to mimic the Fender Blackface preamp. The first stage has low gain to provide about the same headroom as a tube, the diodes on the input clip the incoming signal so as to be pedal friendly and protect the opamp. The diode on the second stage is inverted because the input stage opamp is non-inverting.

To test the circuit out, I used a Peavey Mark III Standard from the 80's. It has two opamp based preamp channels, one with a FMV tone stack, and a handy power amp input on the front panel. The power amp is a 260D (130W at 4 Ohms) with Voltage only feedback. Without the capacitors in series with the JFETs, there was too much subsonic feeding through to the power amp so the caps were added to limit subsonic response.The amp gives a very convincing Blackface tone for both bass and guitar. I especially like it for bass, but the caps could be increased slightly to give a little bit more low end.
Preamps and Effects / Look Ma, NO POTS!
September 25, 2012, 10:10:37 PM
I ran across this circuit on an old Sunn schematic. It's a servo to set the operating point of a JFET so you don't need a pot. Resistors R1 and R2 form a voltage divider, the voltage at the center will be the Drain voltage. The voltage divider will track changes in supply voltage and the opamp will adjust the Source voltage to bring the Drain voltage back.

R3 needs to be low enough that the opamp can still supply the Drain current and not have to swing it's output too close to ground. If the JFET needs a source voltage too close to ground, add a resistor across the 47uF cap, lower than needed to bias the JFET and let the opamp pull it positive to reduce Source current. If you want to add a treble peaking R/C or Fetzer resistor in the Source, add it between the 47uF cap and the Source.

I built this circuit on a proto board with a J201. When I plugged in a MPF102 the bias adjusted itself and the gain was very close to where it was with the J201. The bias adjusts perfectly changing the supply from 9 to 32V. The bias can drift when the output is clipping. You'll get a little jump in bias when the signal cuts off. A slight adjustment to the R1/R2 ratio will fix this.
Amplifier Discussion / Bandwidth of Power Amp
March 15, 2012, 06:27:45 PM
I'm prototyping a solid state power amp. It's class A, sort of a JLH meets Nelson Pass thing. It runs without any compensation although there is a slight oscillation. I added a couple of caps and that killed the oscillation and cleaned up the square wave response. The -3dB bandwidth is still over 500kHz. So I put the typical RC on the input, knocked the bandwidth down to 40kHz. I could go lower, just change one capacitor. I like the single PNP transistor input borrowed from JLH. When you overdrive it, you get a 40/60 square wave.

My question is, how much bandwidth should a power amp have? I know it's important that when you are overdriving it, the slopes should get steeper the harder you drive it. The amp sounds ok to me so far, even without the bandwidth limit.
Tubes and Hybrids / Triode Emulation X-Y plots
December 01, 2011, 08:08:29 PM
This is a different take on the emulating triode tube distortion theme. First, a few acknowledgements are in order. Findeton's thread, is some fine work but I find it a little complicated. Certainly the math gives me a headache. I didn't look at Dimitri Danyuk's paper, I just figured it would be more mind numbing math. The Runoff Groove article is interesting but I need to see waveforms, not just equasions. The one problem I found in the article was the statement "The first valve stage of a Fender amp can withstand input voltages up to +/- 2.5V without noticeable clipping." We shall see that isn't true. Steve Conner's thread is more my style. I know Steve from Ampage, he is very knowledgeabe on tubes and solid state design. I hope to be able to present pictures as good as Steve's

The first method I will present is the opamp with diodes in the feedback path. As I mentioned at the end of Findeton's thread, the clipping stage of the TS-9 does not come close because in the non-inverting configuration, the input signal gets added to the distorted signal due to the +1 term in the gain equasion. That leaves inverting opamps and shunt clipping. A significant problem is that diodes clip at pretty low voltages so, to keep from having to attenuate the signal, you would need lots of diodes in series. I have overcome this obstacle with a circuit I call the Diode Multiplier.

When I was in high school, I was told that a silicon diode drop is 0.6 or 0.7 volts. That's true at around one milliamp. With a 100K input resistor, +/- 2.5V input signal, we would like to know what the voltage drop is at +/- 25 microamperes. I have curve tracer photos of a 1N4148, a germanium diode, and the gate of a J201. The germanium diode is interesting because it starts conducting right at zero volts. The J201 gate stays off for a significantly larger voltage than the 1N4148, that could be useful. Low voltage zener diodes are another surprise. They start conducting around 0.4V and don't reach full zener voltage until the test current of 20mA is reached. The 1N5221B is a 500mW zener rated at 2.4V (at 20mA).

The first circuit uses a TL072. One trimpot sets the gain, the other sets the amount of diode multiplication by attenuating the feedback appled to the diodes. This circuit does not address the grid current of a triode issue, that will be added later. The 10K trimpot was set to divide by roughly 17, 10.26K on the top portion and 724 ohms on the bottom portion.