Velleman used to offer two tube power amplifier kits, the K4040 and K8010. Both are out of production as of now, and even the manuals are no longer available. Some of the web pages that discuss the amplifiers are no longer available, and the information becomes harder to find.
Since I own a K4040, built from a kit back in 2009, and just revised it slightly, I'd like to open this thread to keep the manuals, mods, and other relevant information easily available for reference.
I start with the two original assembly manuals from Velleman. These used to be publicly available on Velleman's web site, so I hope there it's not a problem attaching them to this post.
In the next posts, I will publish the original and modded schematics and mods I have been able to find and use, links to some resources that are still available, as well as some pictures.
Since I own a K4040, built from a kit back in 2009, and just revised it slightly, I'd like to open this thread to keep the manuals, mods, and other relevant information easily available for reference.
I start with the two original assembly manuals from Velleman. These used to be publicly available on Velleman's web site, so I hope there it's not a problem attaching them to this post.
In the next posts, I will publish the original and modded schematics and mods I have been able to find and use, links to some resources that are still available, as well as some pictures.
K4040 schematic and output and power transformers
The schematic for the K4040 is quite conventional: Two triode amplification stages, a cathodyne phase splitter, and a big and hot qual of EL34s in untralinear more driving the output transformer. The feedback is fed into the cathode of the second triode stage; the whole amplifier is inverting.
The output toroidal transformer has Raa of 1800 ohm (15:1 turn ratio between anode-anode primary and 8ohm secondary). The ultralinear taps are 33%. The secondary has a 4ohm tap that is also used for feedback.
The power supply has a whole bunch of nice features not frequently seen in pother designs. It includes:
The schematic for the K4040 is quite conventional: Two triode amplification stages, a cathodyne phase splitter, and a big and hot qual of EL34s in untralinear more driving the output transformer. The feedback is fed into the cathode of the second triode stage; the whole amplifier is inverting.
The output toroidal transformer has Raa of 1800 ohm (15:1 turn ratio between anode-anode primary and 8ohm secondary). The ultralinear taps are 33%. The secondary has a 4ohm tap that is also used for feedback.
The power supply has a whole bunch of nice features not frequently seen in pother designs. It includes:
- A separate standby power supply based on a separate small PC mounted transformer (in the upper left corner of the power supply schematic, marked "TRANSFORMER")
- A thee way switch selecting between OFF (only standby power is live), STB (main power TRANSFO2 transformer is live, heaters are on, but anode supply is reduced) and full ON
- Soft start (Q1)
- Delay between powering on heaters and applying full anode voltage (Q2-Q4); it also controls the muting relays that short the amplifier's input
- Output tube bias meter based around LM3914 and a LED array
Attachments
K4040 IronAudio mods
Over the years, there has been a lot of discussion of what to change in the K4040. One of the more comprehensive lists was posted in 2001 on ironaudio.com, now defunct. The list is still available on web.archive.org (thank you gemenakis for providing the link!), but here it is for your convenience:
Over the years, there has been a lot of discussion of what to change in the K4040. One of the more comprehensive lists was posted in 2001 on ironaudio.com, now defunct. The list is still available on web.archive.org (thank you gemenakis for providing the link!), but here it is for your convenience:
One thing I would add to the list is getting two matched quads of tubes. The original Svetlana EL34s included with the kit were unmatched, and their distortion performance was unpredictable even with properly adjusted bias. Matching (preferably curve matching) each quad is important - some tube stores will do it for free.
Last edited:
Hi alexcp,
it might be of interest that the K4040 had a predecessor called K4000. I built it from a kit about 25 to 30 years ago, still own it and also have the printed manual including building instructions. I remember there was a Dutch website dealing with mods before the year 2000 already. Maybe I have a printout or dump of that site somewhere, too.
IIRC, basically the K4040 is just the K4000 plus an added biasing metering circuit and somefancy chrome rings where the EL34s protrude through the chassis.
Regards, Tom
it might be of interest that the K4040 had a predecessor called K4000. I built it from a kit about 25 to 30 years ago, still own it and also have the printed manual including building instructions. I remember there was a Dutch website dealing with mods before the year 2000 already. Maybe I have a printout or dump of that site somewhere, too.
IIRC, basically the K4040 is just the K4000 plus an added biasing metering circuit and somefancy chrome rings where the EL34s protrude through the chassis.
Regards, Tom
Attachments
Thank you Tom! I remember seeing the manual for the K4000 many years ago. And yours is a good looking amp!
May I ask you to please scan and post the schematic? It is curious, since K4040 is 90W/ch, but the front page of K4000 manual says 2x200W. Does it also use the output tubes in ultralinear or pentode mode?
Update: just found a K4000 manual online and attach it here. It is ultralinear, 200W is “music power”.
May I ask you to please scan and post the schematic? It is curious, since K4040 is 90W/ch, but the front page of K4000 manual says 2x200W. Does it also use the output tubes in ultralinear or pentode mode?
Update: just found a K4000 manual online and attach it here. It is ultralinear, 200W is “music power”.
Last edited:
If an ECC83 cathodyne is enough for this , it means that the rest of phase splitters and more powerfull tubes for drivers are useless 
Last edited:
I had a K4040 changed the el34's for 6550's. My main comment on the original design was the driver stage was very slew rate limited so a 10KHz sinewave at 1/2 power would look more like a triangle wave at the speaker. The first stage also did give heater hum with its un-bypassed cathode resistor. Good kit though.
Thank you for adding Velleman information.
I regard mirror surfaces a design mistake in tube power amps.
Heat radiation is prevented on even two sides of this amp and
heat just "thrown back" to the tubes. As a consequence valve
temperatures are unduly high.
I regard mirror surfaces a design mistake in tube power amps.
Heat radiation is prevented on even two sides of this amp and
heat just "thrown back" to the tubes. As a consequence valve
temperatures are unduly high.
It is a lot of power dissipation. Each tube dissipates 6.3V x 1.5A = 10W on the heater and 380V x 40mA = 15W on anode, 25W in total. Eight tubes in total dissipate 200W, and there is not that much space between them.
Even more radical mods
In 2006, ecsdesign published here on diyAudio a much more radical way to modify the K4040, which I ended up implementing. His original posts are here and here - if the links take you to a wrong place, they are posts #1029 and #1037 on the thread "Building the ultimate NOS DAC using TDA1541A".
For convenience, here is his first post (the second post will follow):
In 2006, ecsdesign published here on diyAudio a much more radical way to modify the K4040, which I ended up implementing. His original posts are here and here - if the links take you to a wrong place, they are posts #1029 and #1037 on the thread "Building the ultimate NOS DAC using TDA1541A".
For convenience, here is his first post (the second post will follow):
Last edited:
If this schematic is correct an EL34 tube will go full current when the bias pot looses contact.
Cathode resistor will be destroyed and probably screen resistor also.
It should be possible to find a quick fix for it.
Cathode resistor will be destroyed and probably screen resistor also.
It should be possible to find a quick fix for it.
The schematic is correct. A fix is probably possible but not without modifying the board. Using better quality trimpots should help.
Fix : jumper the wipers to the -VL end of the pot, probably decrease
the series resistor by a parallel one if the adjustment range suffers.
Edit : additional resistors in series from -VL to pot are required also.
the series resistor by a parallel one if the adjustment range suffers.
Edit : additional resistors in series from -VL to pot are required also.
Last edited:
ecdesigns's modification removes the global feedback from K4040. Without feedback, K4040 has a higher input sensitivity, more hum and noise, and depends on matching the output tubes between the channels.
There are two ways to re-introduce global feedback after ecdesigns's mod:
I tested both and ended up implementing the second option. Attached is the final schematic. CF (47pF) helps to improve the phase margin. CC (22pF) was required for stability of the left channel, which is connected to V11 via a long trace across the PCB, with two wire jumpers.
With feedback, the distortion and hum are reduced, and the sensitivity in in line with the sources like CD players, giving 2Vrms full scale.
There are two ways to re-introduce global feedback after ecdesigns's mod:
- Apply it to the grid of V9B (see the schematic attached to post #12) by lifting the bottom end of CA from the ground and connecting it to the feedback divider. This option is easier to wire and has less stability issues, but it does not enclose V11 in the feedback loop and, as the sensitivity of the rest of the amplifier decreases, makes V11 work harder, increasing distortion.
- Apply it to the cathodes of V11. This is trickier to introduce from the stability viewpoint but improves overall linearity.
I tested both and ended up implementing the second option. Attached is the final schematic. CF (47pF) helps to improve the phase margin. CC (22pF) was required for stability of the left channel, which is connected to V11 via a long trace across the PCB, with two wire jumpers.
With feedback, the distortion and hum are reduced, and the sensitivity in in line with the sources like CD players, giving 2Vrms full scale.
Attachments
In 2005, Jan Veiset published a different way to modify the phase splitter. Unfortunately, his website (www.veiset.net) is gone. A copy is available on web.archive.com, but it is in Norwegian with black letters on black background 😱
Here is a slightly edited Google translation:
Original circuit description
The phase inverter is built around the double triode ECC83. The first half is used exclusively for gain, just under + 40dBu, and the other half is a split-load phase inverter. The advantage of split-load is that it is a simple and tidy construction, and they are easy to balance. The disadvantage is the lack of the ability to deliver (high) voltage.
A slightly more technical review
The triode that works as an amplifier has the following working points: Ia = 0.65mA, Ug = 0.65v (blue curve). The stage works with a relatively high anode load, 390k and the input impedance to the next step is very high, partly due to positive feedback. All in all, the stage works atop maximize the gain.
The next stage is a little harder to understand. It is basically a completely traditional split-load phase inverter, but without the negative bias on the grid! Although the grid leakage is not very high (820k), it is possible that negative bias builds up when the signal is applied. In any case, it looks like this half of the tube is running very hard. Which is also necessary to be able to deliver enough voltage to the output stage.
For driving of the power stage (4xEL34, B+ 420v) approx. 25 volts RMS is needed. If we add 50% to the headroom, the phase inverter should be able to deliver around 37 volts RMS or 100 volts p-p. Both theory and measurements show that the stage has very small margins. The phase inverter is and will be a bottleneck in this construction.
New phase inverter
To avoid too many changes, I keep ECC83, move away from the split-load inverter and replace it with a paraphase splitter. The static working point is the same for both triodes, Ia = 1.4mA, Ug = 1.4v (red curve). I lose some gain, estimated gain with R1 in the next stage around 130k is about + 33dBu or a loss of about 6dB compared to the original solution. Umax is designed for 40v RMS with Ub = 365v.
At the same time I change the coupling capacitors from 22nF to 220nF. This will lower the open loop frequency response from around -3dB @ 30Hz down to -3dB @ 3Hz. Hopefully this gives better control in the bass area.
Practical implementation
The practical implementation was easy, as I had prepared by drawing the new components on a copy of the top of the PCB.
After some soldering work, it was time for measurements. The result was above all expectations, about 55v rms before phase splitter clipping. That is, there is plenty of headroom here.
Picture before and after are attached.
Working drawing
Here is the working drawing I worked from. If you want to upgrade the phase inverter in your K4040, it should work well if you follow the following recipe:
1) Remove the following components:
J23
R86, R87, R88, R23
C15, C19, C20, C21, C22
2) Cut the PCB trace between R87 and C19 / C20
3) Fit the following components:
Rg (680k)
R86, R88 (100k)
R87 (1k)
Rf1 (593k 560k + 33k)
Rf2 (560k)
C19, C20, C21, C22 (220nF)
Cf (100nF)
Ck (100uF)
4) Connect R88 and C19 / C20 on the underside of the PCB (yellow line)
5) Do the same for the other channel.
6) Screw the amplifier together and turn up the volume.
Here is a slightly edited Google translation:
Original circuit description
The phase inverter is built around the double triode ECC83. The first half is used exclusively for gain, just under + 40dBu, and the other half is a split-load phase inverter. The advantage of split-load is that it is a simple and tidy construction, and they are easy to balance. The disadvantage is the lack of the ability to deliver (high) voltage.
A slightly more technical review
The triode that works as an amplifier has the following working points: Ia = 0.65mA, Ug = 0.65v (blue curve). The stage works with a relatively high anode load, 390k and the input impedance to the next step is very high, partly due to positive feedback. All in all, the stage works atop maximize the gain.
The next stage is a little harder to understand. It is basically a completely traditional split-load phase inverter, but without the negative bias on the grid! Although the grid leakage is not very high (820k), it is possible that negative bias builds up when the signal is applied. In any case, it looks like this half of the tube is running very hard. Which is also necessary to be able to deliver enough voltage to the output stage.
For driving of the power stage (4xEL34, B+ 420v) approx. 25 volts RMS is needed. If we add 50% to the headroom, the phase inverter should be able to deliver around 37 volts RMS or 100 volts p-p. Both theory and measurements show that the stage has very small margins. The phase inverter is and will be a bottleneck in this construction.
New phase inverter
To avoid too many changes, I keep ECC83, move away from the split-load inverter and replace it with a paraphase splitter. The static working point is the same for both triodes, Ia = 1.4mA, Ug = 1.4v (red curve). I lose some gain, estimated gain with R1 in the next stage around 130k is about + 33dBu or a loss of about 6dB compared to the original solution. Umax is designed for 40v RMS with Ub = 365v.
At the same time I change the coupling capacitors from 22nF to 220nF. This will lower the open loop frequency response from around -3dB @ 30Hz down to -3dB @ 3Hz. Hopefully this gives better control in the bass area.
Practical implementation
The practical implementation was easy, as I had prepared by drawing the new components on a copy of the top of the PCB.
After some soldering work, it was time for measurements. The result was above all expectations, about 55v rms before phase splitter clipping. That is, there is plenty of headroom here.
Picture before and after are attached.
Working drawing
Here is the working drawing I worked from. If you want to upgrade the phase inverter in your K4040, it should work well if you follow the following recipe:
1) Remove the following components:
J23
R86, R87, R88, R23
C15, C19, C20, C21, C22
2) Cut the PCB trace between R87 and C19 / C20
3) Fit the following components:
Rg (680k)
R86, R88 (100k)
R87 (1k)
Rf1 (593k 560k + 33k)
Rf2 (560k)
C19, C20, C21, C22 (220nF)
Cf (100nF)
Ck (100uF)
4) Connect R88 and C19 / C20 on the underside of the PCB (yellow line)
5) Do the same for the other channel.
6) Screw the amplifier together and turn up the volume.
Attachments
thanks so much to alexcp for all the information he provides on this protentional great sounding amp... and for being a gentleman!
Thanks a lot!
Thanks a lot!
