I have built this amplifier twice, first time when I was 16 years, the second time two years later and that was a pretty nicely finished build in a 19 inch box, I later sold that amp and regretted it. In the second build i used TIP35C as output transistors all the other components were according to the schematic. I had an oscilloscope the second time and and I tortured the poor amp with sine bursts from a burst generator i built, also from Elektor magazine. When I think back I am still surprised how well behaved the amp was, there was no instability, it clipped cleanly with quick recovery, I tried low loads 2ohms and capacitive load 2.2µF, it never broke despite me disconnecting the diodes and squeezing 20A peak out of it. The bias was stable and i tried raising it to 100mA because then I thought higher is always better.
Publish and be damned 40years later, this design is from 1982 and I think for the time it was pretty good it also it uses cheap and still available components. I have contemplated building it again for nostalgic reasons and it sounded good to. I brought the second build to a friend who just bought a Creek amplifier and Heybrook HB1 speakers, my amplifier decidedly beat the Creek amp and my friend had to begrudgingly admit defeat.
Publish and be damned 40years later, this design is from 1982 and I think for the time it was pretty good it also it uses cheap and still available components. I have contemplated building it again for nostalgic reasons and it sounded good to. I brought the second build to a friend who just bought a Creek amplifier and Heybrook HB1 speakers, my amplifier decidedly beat the Creek amp and my friend had to begrudgingly admit defeat.
R4 is high value 4.7K so total needed capacitance is lower for the high pass filter it forms.
Somewhat a neat trick, to not have to use large value polarized caps.
Use small non polarized. 4.7k / 2.4uF be about 14 Hz cutoff frequency.
Amplifier has a interesting way of getting better gain for the power devices
and still use all NPN outputs. pretty cool.
uses transistors as diodes also interesting.
If it thermally behaves like CFP the small TO-92 transistors.
could just be placed close together or even mounted flat to flat
and thermal tracking all done.
whenever a simplified symmetrical input/vas design gets posted.
people will whine about it. simplified still works.
vas current dictated by input current. hence why constant
current sources. both rails, possibly better noise rejection
less distortion. still simple. better ways to do it of course.
just add 20 more parts.
CCS with higher voltage ref like zener 5.6v even 8.2 volts.
has good output impedance like feedback ccs. removes
extra transistors, requires higher resistor values.
noise rejection is better, less parts. resistor values
dont cook so much.
Basically just an old BD & 2N3055 classic.
Nothing really special, but a true 'work-hoarse' of the day.
Nothing really special, but a true 'work-hoarse' of the day.
I once did an amp shootout/compare between my 2x EL84 ultralinear and an Elektor amp. the poor owner found out that his record player had a lot of rumble when connected to the tube amp, as his KEF104 passive cone was moving like crazy on the tube amp. The elektor amp did not go that low. Philips ultralinear output transformer..
Yes in the original article the tell you to thermally couple T9-T11 and T10-T12.
You could increase R8 and it would still work but would it be good? As it is now the zener current is around 6mA with +-35V supply and if you look in the datasheet for the old classic Vishay BZX55 zeners the zener impedance is max 25ohm at 5mA if you decrease the current to 1mA the zener impedance goes up to 450ohms so the higher current gives higher stability, and probably less noise to. As others have commented the VAS current is dependent of the LTP current so a nice and stable LTP current would be good, do you still want to increase R8?
Everyone is use to higher values for feedback CCS
seen dumber things like using less predictable leds
or using extra parts like 2 diodes.
one zener, all done less parts, better noise rejection.
seen dumber things like using less predictable leds
or using extra parts like 2 diodes.
one zener, all done less parts, better noise rejection.
you turn on the amp, the internal temp is 18C, after 1hr the internal temp has risen to is 38C, consider what that does to your ccs output.
Last edited:
5.6V zeners have very low temperature coefficient, among the lowest you can get without going to exotic reference circuits, you still have the transistor VBE that will vary with temperature but since the voltage drop over the emitter resistor is around 5.0V, the voltage change caused by VBE change results in proportionally less current change than if you have a very low reference voltage like two diode drops. This is a pretty temperature stable CCS.
Note the tempco is -0.05 to 0,05 for the 5.6V zener, it could actually be zero if you are lucky.....
Note the tempco is -0.05 to 0,05 for the 5.6V zener, it could actually be zero if you are lucky.....
Very stable.
and adding another higher value zener
or comically a current source for your current source.
then the rail voltage can change dramatically
drop/rise and current always the same.
using higher voltage ref for ccs is better performance.
5.6 even 8.2 work very well.
tested it before in many circuits
at 18c current output is 1.1ma and 38c is 1.1ma
to be exact at 18c =1.12 ma and 100c 1.14 ma
current remains constant.
Unless only the zener got very very hot.
current will slightly fall. and transistor was magically very very cold.
even then constant current doesnt change much
otherwise transistor be hotter and does fine up to 100c even higher.
low voltage like switching diodes even 2 volt led
with less predictable current. waste of time
and burns up low value resistors.
1zener 1 transistor, less parts better temp co
dont oscillate. same old output impedance.
reject more noise
and adding another higher value zener
or comically a current source for your current source.
then the rail voltage can change dramatically
drop/rise and current always the same.
using higher voltage ref for ccs is better performance.
5.6 even 8.2 work very well.
tested it before in many circuits
at 18c current output is 1.1ma and 38c is 1.1ma
to be exact at 18c =1.12 ma and 100c 1.14 ma
current remains constant.
Unless only the zener got very very hot.
current will slightly fall. and transistor was magically very very cold.
even then constant current doesnt change much
otherwise transistor be hotter and does fine up to 100c even higher.
low voltage like switching diodes even 2 volt led
with less predictable current. waste of time
and burns up low value resistors.
1zener 1 transistor, less parts better temp co
dont oscillate. same old output impedance.
reject more noise
Last edited:
theory and simulation of a sub-circuit is one thing.
real measurements taken on a fully assembled amplifier are another.
real measurements taken on a fully assembled amplifier are another.
Although I share many of Turbowatch' experiences with oscillating Elektor/Elektuur amps (the "Ekwin" Quad 303 implementation being one of them) the circuit shown here worked flawlessly from the start. Several readers here - Ian Finch i.a.- have built this particular amp with more than positive experiences too; one member turned it into a fully complementary output stage with MJE 2955/3055.
Some discussion here: https://www.diyaudio.com/community/threads/2n3055-mj2955-amplifier-kit.298832/
https://www.diyaudio.com/community/threads/amplifier-based-on-2n3055.10337/page-4
Some discussion here: https://www.diyaudio.com/community/threads/2n3055-mj2955-amplifier-kit.298832/
https://www.diyaudio.com/community/threads/amplifier-based-on-2n3055.10337/page-4
Yes, because if the zener impedance increase to 450R so does R8 to 68K, so for stability it s like a 6k8 feeding a zener that has 45R impedance as far as we re talking of the power supply voltage variation, comparatively to 25R that s not a big change, and R8 at 68K dissipate about 0.06W instead of 0.6W for 6k8, that s not all to have a higher current if components are to be thermally tortured, long term reliability is to be a concern in any design, not counting efficency.
How do you get 0.6W in the 6k8 resistor? I get 0.2448W, if the total supply voltage is 70V how high is the zener current? First we lose 11,2V total over the zeners themselves we have 58.8V left, the current is limited by R8+R11+R12 that is 9800ohm resulting in 6mA. that results in 40.8V over R8 and 0.2448W. We lose a little more over R11 and R12 because of the 0.5mA going through then to the LTP but i ignore that. The zeners will dissipate 33,6mW.
The only components getting a bit toasty in this design is T7 T8, at 8mA and almost 34V over them that is 270mW or so in a TO92, I remember them getting a bit warm in my build.
The only components getting a bit toasty in this design is T7 T8, at 8mA and almost 34V over them that is 270mW or so in a TO92, I remember them getting a bit warm in my build.
Because I was previously working on shunt regulators
and my train of thought was backwards.
Good point with zener bias current source seems dissipation
in some resistors would actually be higher.
around 1mw compared to 10mw
looking at basic benefits/ tradeoffs
Zener bias for current source is more stable over wide temperature range.
And all in all resistor degen values used, will bring output impedance
to around 2Meg.
Common circuits using dual switching diodes or even Led, emitter degen
wont be as high. Output impedance maybe 300 to 750K
Tradeoff with zener or higher voltage bias. There is overall voltage
losses , less swing or overhead.
So not a good idea for a VAS stage or gain stages for max swing to rail.
Feedback current source more beneficial and impedance about the same.
Zener more useful for input differential, has high impedance and
more stable over temperature range. So DC offset for a amplifier
in theory wont drift as much. And if rail voltage sags.
Bias voltage still remains high enough to hold output impedance to 2Meg
With feedback current source if rail voltage sags then impedance also goes down.
Theoretically should be about 2Meg, but can drop to 1Meg
And current changes with higher temps.
always trade offs. Less loss of overhead, good for gain stage if your trying
for maximum swing to rail
Otherwise very simple Zener's more thermally stable, source impedance doesnt drift as much.
Ideal for input differential.
Hence comment for this amp and common issues with simple symmetrical amplifiers.
Best thing to use and used, be zener current source. very stable over temp range.
important since differential dictates all the other stages.
Last edited:
small wonder with some 8-12 dB bass boost because of current drive of the tube amp instead of voltage drive of the solid state amp?