It might be oscillating. The bias does jump when the probe is touched to some points. Also, for these last sets of readings I plugged direct to the wall socket, not through the variac so the voltage is a little higher. In the morning I will pull Q25 & 26 and see if that changes anything. There has to be a reason why Q8, 10 &12 are not turning on.
Thanks, Terry
Thanks, Terry
Hi Guys
Looking at the voltages on Terry's schemo, the input circuit seems to be pulling a lot of base.current. The base of Q1 should be nominally at zero volts, but (although it is not indicated directly) it is inferred to be 1.58 - 0.403 = 1.177.
The feedback node (base of Q2) is at 1.15V, so the difference voltage is 27mV, with the input positive compared to feedback.
1.177V across 47k5 suggests about 25uA of base current. If a similar current was flowing through R16 as it should be, the output offset would be just 10s of mV. Considering that 3mA should flow through Q1 and Q2 at idle, 25uA of base current suggests a healthy beta for Q1 or that it is pulling all the current.
Q12 should only turn on if the current through Q8 is excessive.
Note also that 2SC3503 is a better choice for Q8, given that its base-collector capacitance is one-tenth that of 2SC4793. THD will be lower with the 3503, too. Remember to stuff the TO-126 backwards to the silk-screening.
At the other end of the circuit, the protection transistors (25,26) seem to be on hard. Try lifting one end of R54 and R55 and see if everything corrects. If it does, leave out these two Rs.
The voltage across the shunt feedback cap (C4) is clamped by the diodes to about 1.4V either direction. This makes me suspect that the "ground" used for the meter reference is not really ground. Use a circuit ground not the chassis.
Have fun
Kevin O'Connor
Looking at the voltages on Terry's schemo, the input circuit seems to be pulling a lot of base.current. The base of Q1 should be nominally at zero volts, but (although it is not indicated directly) it is inferred to be 1.58 - 0.403 = 1.177.
The feedback node (base of Q2) is at 1.15V, so the difference voltage is 27mV, with the input positive compared to feedback.
1.177V across 47k5 suggests about 25uA of base current. If a similar current was flowing through R16 as it should be, the output offset would be just 10s of mV. Considering that 3mA should flow through Q1 and Q2 at idle, 25uA of base current suggests a healthy beta for Q1 or that it is pulling all the current.
Q12 should only turn on if the current through Q8 is excessive.
Note also that 2SC3503 is a better choice for Q8, given that its base-collector capacitance is one-tenth that of 2SC4793. THD will be lower with the 3503, too. Remember to stuff the TO-126 backwards to the silk-screening.
At the other end of the circuit, the protection transistors (25,26) seem to be on hard. Try lifting one end of R54 and R55 and see if everything corrects. If it does, leave out these two Rs.
The voltage across the shunt feedback cap (C4) is clamped by the diodes to about 1.4V either direction. This makes me suspect that the "ground" used for the meter reference is not really ground. Use a circuit ground not the chassis.
Have fun
Kevin O'Connor
Last edited:
Yes, input is and has been shorted since the start.
I am trying to have fun. 😉
So, looking at my schematic, Q1, Q8, Q10, Q12, Q13, Q14, Q15, Q16, Q25 and Q26 are not conducting. I have checked every resistor and they are correct according to the schematic. Can someone please sim this amp and see if the circuit works as drawn before I waste any more parts? I shared my extra boards so others are hoping to build this amp too. I sure would like to gt it working so they will have a base to work from.
Thanks, Terry
Terry,
I can see about redrawing it in LTSpice to see if the published design at least simulates alright over the weekend. I wish I had all the parts on hand to get a build to where I can verify wether there is a flaw in the design / layout, or with your specific build and offer some assistance, but that is going to be a little while yet.
My personal modus operandi is to always build at least one of something before letting anyone else have it. That way, should it not work for someone, I can be sure there is a faulty part or an issue with that individual's build and not an issue with the design or layout. I'm curious, was this amplifier ever built by its author or anyone else as validation prior to release?
I can see about redrawing it in LTSpice to see if the published design at least simulates alright over the weekend. I wish I had all the parts on hand to get a build to where I can verify wether there is a flaw in the design / layout, or with your specific build and offer some assistance, but that is going to be a little while yet.
My personal modus operandi is to always build at least one of something before letting anyone else have it. That way, should it not work for someone, I can be sure there is a faulty part or an issue with that individual's build and not an issue with the design or layout. I'm curious, was this amplifier ever built by its author or anyone else as validation prior to release?
Terry,
I can see about redrawing it in LTSpice to see if the published design at least simulates alright over the weekend. I wish I had all the parts on hand to get a build to where I can verify wether there is a flaw in the design / layout, or with your specific build and offer some assistance, but that is going to be a little while yet.
My personal modus operandi is to always build at least one of something before letting anyone else have it. That way, should it not work for someone, I can be sure there is a faulty part or an issue with that individual's build and not an issue with the design or layout. I'm curious, was this amplifier ever built by its author or anyone else as validation prior to release?
That is what I am doing in my "triple thread" ... Bimo and Naf already built
the EF3 ...
Now (and only now) would I consider "group buy" or to release to the masses.
The HB was put down on this thread ... but it's flaws are known and worked out.
I did not comment on this EF3 , as it has several shortcomings (where o'
where are those OP basestoppers ?) Triples are "picky" 😀 .
OS
Hi Guys
There is nothing unusual about the circuit. It is all standard stuff that Self simmed and built, and that I have built over the years. I don't do sims - just build and test. Yes, reality is quite different than virtual...
The only thing different in this amp than what I have built is the cascoded front-end - I've never bothered with cascodes. That should not keep it from working except at low voltage. I've never had any trouble with triples - or tribbles. Maybe I am charmed when I build stuff since any amp I've built has worked first time and not failed? Could give me a false sense that circuits work? I've never gotten a simulator to work properly, but they don't make any for linux...
As long as there are NPNs and PNPs in the proper places it should all function. Using MJEs for Q8,7 should not effect functionality. However, they may be over-dissipated at full voltage with 7mA or so through them.
I would clip the meter to a ground point on the board rather than a distant star. It still seems like there is a problem with the grounding.
Have fun
Kevin O'Connor
There is nothing unusual about the circuit. It is all standard stuff that Self simmed and built, and that I have built over the years. I don't do sims - just build and test. Yes, reality is quite different than virtual...
The only thing different in this amp than what I have built is the cascoded front-end - I've never bothered with cascodes. That should not keep it from working except at low voltage. I've never had any trouble with triples - or tribbles. Maybe I am charmed when I build stuff since any amp I've built has worked first time and not failed? Could give me a false sense that circuits work? I've never gotten a simulator to work properly, but they don't make any for linux...
As long as there are NPNs and PNPs in the proper places it should all function. Using MJEs for Q8,7 should not effect functionality. However, they may be over-dissipated at full voltage with 7mA or so through them.
I would clip the meter to a ground point on the board rather than a distant star. It still seems like there is a problem with the grounding.
Have fun
Kevin O'Connor
Last edited:
for my builds, output off-sets not more than +-100 mV,
and that output stage idle current is adjustable via the pot...
if i have both of these two, then its play time.....
and that output stage idle current is adjustable via the pot...
if i have both of these two, then its play time.....
I'm trying to build a spice model for this amp. Tired of wasting parts. Can someone tell me how to find a trimpot component in LT spice?
Thanks, Terry
Thanks, Terry
There isn't a trim pot. Just use a regular resistor and rename it. You can set the value to whatever you need it to be. I have a good amount of the circuit already entered into LT Spice so hopefully I can get the circuit verified and put something up to check the build against.
Thanks Jason. I'm going to keep going because I want to learn how to use it. Just not sure how to wire up a resistor with three poles. I may have to wait to see how you do it.
Blessings
Blessings
Look at how the resistor is used in the circuit. It has the wiper shorted to one end, so it is purely a variable two leaded resistor. There are tricks to make a potentiometer in LTSpice but not needed here for this purpose.
There are some good models at Cordell Audio. Other members may have collections available as well. Otherwise you can often get models from the manufacturer, though these often can be off in terms of accuracy.
I have built a model in LTspice but I have no idea how to run it or if it is even correct. Perhaps one of you could check it out for me. I basically used the same transistors in the model that I used in this amp, I think.
Hi Tony,
I have one pair left but need them to help search part numbers and values because the resistors cover the numbers. Once I get the amp working, you can buy them if you want.
Blessings, Terry
I had to fix some errors and put in device models from my library, but it runs.
Nowhere near optimized in terms of the simulation, just a quick 'n dirty pass.
May have spoke too soon, noticed no load. With load the sim is taking a looong time, likely oscillating in spice.
Nowhere near optimized in terms of the simulation, just a quick 'n dirty pass.
May have spoke too soon, noticed no load. With load the sim is taking a looong time, likely oscillating in spice.
Last edited:
It won't run on mine because I don't have the models you used. Does it run well enough to check the voltages against my schematic? I would really like to find where my amp is broken.
I'll have to go through it to ensure there's no other errors or missing components and put in the appropriate directives to get closer to reality. It is getting later in the evening where I'm at, so it will likely be sometime tomorrow before I can get a better pass. When I have some voltages to post from the sim I will do so.
- Status
- Not open for further replies.