Laney Klipp 100W Repair and Design Study – Part 2
I found a couple of small schematic errors that had me head scratching a bit trying to follow the Normal and Klipp channel outputs. This was a presumption of pin numbers at valves 1 and 2, which I have rectified by re-connecting them correctly in the diagrams in this section, as well as the PT bias diode being reversed in the schematic.
As seen on the video, the amp is working with just some minor issues like a scratchy presence pot – a new cap there should help to keep the DC off it etc. – and a loud mains hum on the Klipp Hi Z input only, when plugging a lead in here.
I want to document some aspects for future reference such as circuit voltages, power output, scope signal shape etc.
I took readings for all valve anode and cathode bias voltages once it was running on the mains.
I got to that stage by powering up the amp with the Variac, gently in stages. As I intend to test gently I can use the incorrect value but correct size 3A fuses from the Carlsbro amps as a temporary measure until new 32mm x 6mm fuses arrive.
The first tests are always with valves removed so no large current can flow into a completely cold and possibly very long time unused amp.
I tested the power transformer with no standby on. I wound this up slowly in steps – 50V, 80V, 120V, 150V, 180V, 200V then 240V, watching for smoke and hot smells from the PT. This also puts voltage across the secondaries and heaters windings, but with no current load.
This is done so that the insulation and wires can warm up a little from a small magnetizing current in an unloaded transformer, and checks the insulation breakdown potential should arcing to the chassis or between windings occur like the JMP did. If that happens inside the PT then the wires will probably short then melt and that will be the end of that.
Once I have powered up to 240V with the Variac a couple of times, I know the PT can handle the voltage stress at least if not the current to come yet.
I also find out what voltage these old neons start to light at (about 100Vac) as a future indicator.
When confident all seems good I then do the same again but with the standby switch on to power up the rectifiers and DC section. I use a scope probe at A+ to see rectification occurring, as well as a meter on the 1000V DC range.
Once the amp is taken to maximum 240V, I know roughly what the DC HT voltage I will be dealing with in future (about 660V) and that if anything is going to fail at these higher voltages it should be now – not later with the valves in that could mean damaging valves as well as a PT and/or OT.
Once I have cooked the amp a bit and switched up and down with the Variac a few times I can turn off and replace all the valves and do the same slow stage tests again, because the heater windings have to be tested under load this time separately from the rectifier/DC section so I know which section has a fault if one occurred.
I am paying more attention for smoke and heat with the valves in but all knobs at zero (0 volume as there is no speaker attached yet), ready to turn off asap to minimise potential damage due to a fault.
I check the value up to full HT (565V) for mains ripple at A+ to D+ at this point too with each switch on and off, as ripple rises markedly with valves in than out (7Vpp-40Vpp in this case).
Above – 40Vpp A+ ripple at full HT 565V
I like this amp as it sits in a stable position because of the large Partridge transformers when upside down, and this also means the EL34s just fit underneath without being rested on very much. I put an aluminium sheet under these to stop the heat to the wood bench:
Once happy all is well I can remove the Variac and just use the mains direct once a speaker is attached, and test for sound by turning up the volume gently to hear any mains hum or hiss, then power off.
If that went ok, I usually just want to hear it, so can plug in a guitar and play a bit – being VERY careful with the underside exposed at 565V.
This will indicate any other bad noises obviously. If all is good, then I can start taking circuit measurements like HT, all valve anode and cathode voltages and signal tests with FLS if I wish, like the end to end signal distortion from sine wave in to speaker jack out.
I marked the voltages I found on the schematic and finally ironed out for myself what exactly goes on with the bias diode on the tertiary PT winding; 75Vac is ½ wave rectified to create a negative (about -50V) rail for biasing the power valves Vgk:
This can be understood now with the diode the right way round, as the when the 75Vac winding goes positive at the diode, no current can flow, so there is 0V DC from the connection to ground. When it goes negative, the diode conducts creating -75V DC at the diode and 15k resistor. The AC is rectified by the 56k iin parallel with the 15uF electrolytic cap, with its positive at 0V DC, and its negative at about -50V DC. This is connected as below to the anode side of the valve 4 LTP.
Below, note the 2 x 100nF, 400V DC blocking and signal AC coupling caps, that separate the 304V HT for the LTP and the -47V HT that biases the EL34 grids/cathodes and feeds the signal to them. For a 200mV AC input to the LTP signal grid (38V point) I got 1V AC at the anode – a gain of 5.
This 1V to the grid inputs to the EL34s that gave a 20V anode output – a gain of 20 for the power valves.
I have corrected the channels that actually connect to the anodes of valve 1 in my schematic below. The Klipp channel connects to pin2 grid in the real amp.
For a 200mV input at the Normal channel (red), I got a pin6, 12V anode output, a gain of 60.
For a 200mV input at the Klipp channel (blue), I got a pin1, 10V anode output, a gain of 50. This then feeds the grid of pin7 (also reversed from my presumption in part 1) of the LTP valve 2. This gives an output at the pin1 anode of 5Vpp clipped pulse wave as below which generates the Klipp channel distortion signal:
This can be switched in or out and mixed with the bypassed Klipp clean signal at the bottom of the 1M mix pot as desired before passing via the 1M volume to the input grid of valve3, the cathode follower.
I am now in a position to re-draw the Load Lines with actual values for this stage if I like, to understand how and why the pulse clipping is generated here.
The main thing is to see any voltages that are wildly out from a sensible working value that indicate incorrect settings such as bias voltages that are way off for the type of amplifier. I can see that valves 1 and 2 look fine, but I am not familiar with the DC coupled cathode follower of valve 3, where the output of V3 triode1 connects directly to the input grid of V3 triode2. I have to research this to know that the 161V pin 8 cathode voltage is “normal” even though all is working.
Seems all these values are near identical as in Blencowe’s example above, and book p122.
I was curious how the input and output for this valve stage works – all I knew was it should be unity or less, and its output impedance matches the tone stack that follows. This appears to be a variation of a classic Fender-Marshall-Vox (FMV) tone stack on page 207 of Blencowe, but not sure.
After taking some measurements at valve 3 by following the Normal channel red path at 200mV input, it was bit clearer what happens here. I increased the Normal volume from 0 until I got 200mV at the pin2 grid of valve3. I measured nothing at the cathode of pin 3 above the 820R. I got 200mV at the cross connected anode/grid at pins 1 and 7, and 200mV at the output cathode pin 8.
This shows that there is indeed a 1:1 unity gain, so this valve acts as a Push Pull stage because, as the input grid goes positive then the anode goes negative by the same value due to the 100k load resistor and a gain of 1. As this anode goes negative, then so does the other grid as they are connected, so the second triode goes toward shut off with less current flowing. This means the output cathode also goes negative as less current flows through the 100k cathode resistor, so this whole valve stage is an inverter of the grid input – Blencowe agrees with me:
As both triodes have a 100k resistor in the circuit (except for the negligible 820R) symmetrically opposed (an anode load and a cathode load) then the maximum current that can flow in each side is the same, as the same HT is shared, so the voltage swing across each triode has to be equal but opposite also, hence a unity gain.
Valve 4 LTP
As I know the output from valve 3 is Unity from either Klipp or Normal volumes 0 to maximum, then passes to the tone stack, then whatever exits there – if all the tone pots are full up – is the maximum signal that can go to the pin2 grid of V4. With all tone knobs set flat (mid position), using the Normal volume I set the input of the V4 LTP grid to 200mV and got 1V at the anode. This gives a gain of 5 for V4.
The 1V from above is the input at the power valves grid, which gave a 20V signal at the anodes of the EL34s. I checked the symmetry for each Push Pull side of the anodes, which was made more difficult because of the 40Vpp ripple that it is superimposed on:
You can also see the 180 degree phase between the two halves of the EL34 output sides in these pictures.
Another thing to notice is there are exactly 10 mini sines of the 1kHz test tone per 1 x 100Hz ripple cycles, showing that each ripple peak is twice UK mains 50Hz cycle (count them!)
The Push Pull symmetry seems good, but this would need to be seen at higher volumes to see major differences.
The grid curves for an EL34 are:
The maximum bias is only -25V, so this amp is way beyond that at -47V which would suggest they are permanently cut off?!
Not sure about this so will have to research more.
(A quick googling – seems that the grid bias voltage is not so relevant because these 100W amps HT are beyond the spec. sheet so it is plate current and therefore maximum plate dissipation power that matters to not burn out the valve. This means the negative bias spec. sheet figure would “appear” to cut the bias current off, but the tube is actually at a much greater voltage, so still works as the grid curve is in a different place at these HT voltages – provided the anode current is between the hot and cold values on the chart at
it would be ok).
Now it makes sense, because if you do the maths for an HT of 565V which this amp has, and 25W max dissipation per tube, you get 25W/565V = 44mA of plate current, which is still mid range for the tube tolerance overall.
16 ohm Speaker Output
As I had 20V at the anode, I measured 2V at the speaker for this volume level, which indicates a 10:1 step down ratio for the 16 ohm winding. The table from Briggs below implies then that a 16 (15) ohm coil at a 10:1 ratio would be an OT primary impedance of 1500 ohms.
Hmm, this could be noisy or not possible to do as I dont have large load resistor to replace a speaker. Ideally I have to have the volume at maximum without any distortion at any stage, from a clean channel (Klipp clean or Normal) and measure the voltage at the speaker (16 ohm tap in this case). From P=V squared / R, the actual power out can be calculated.
I managed to stand a fair bit of the near full volume of the Normal channel for which I used the DDM on the RMS AC range, to a max of 40V. From above, if the 16 ohm secondary has a 10:1 ratio with the primary, it would mean the primary is swinging to 400V RMS. This would be 800V peak to peak RMS from each side of the push pull extremes, or 800 x 1.414 = 1130Vpp or more at full volume.
This corresponds to 100W to the speaker, but I’m not convinced as the Marshall Marquee is rated at 66W but didn’t distort at this volume, which wasn’t quite full volume either, but I didn’t push it.
P = V^2/ R = 40^2/ 16 ohm = 1600 /16 = 100W
So, unless the amp can give more than 100W this may be wrong as it didn’t sound loud enough after hearing the Marshall JMP at full (deafening!). This was only a sine wave though – a full guitar would sound very different.
I’ll tell you when I have the new 50 and 60W, 8 ohm speakers I got for the Carlsbro cab which I will wire in series to make 16 ohms, handling 110W. I will have a direct comparison with the 60W Carlsbro then also.
Again, I couldn’t measure at full volume but it looks like this will give out 6Vpp or more from the grid of the V5 LTP. This is 3V peak, or 3 times line level which is about 1V peak or 0.707V RMS. 0.775V RMS = 0dBU.
This would drop a bit under load when connected to a mixer etc. so is possibly about right.
An Excel Frequency Response graph plotted, but at low volumes only. I am only interested in the overall response pattern to get an idea of the fequency output gains of this amp.
Hmm, looks more like an amp with no NFB.
This is a quite a low value as the NFB link from the 16 ohm tap is split 100k:5k on the left – taken above the presence pot.
This 20:1 ratio, or more accurately (but not necessary as 1/20 will do), as the total NFB load is 100k + 5k = 105k = 100% of the load, then 5k is 4.8% of the total. For every 1V at the speaker, 0.048V are fed back to the grid, so:
This corresponds to an overall NFB of about 20 x log 0.048 = -26dB. Very low indeed, so makes sense. This is not the point of theses resistors as such though, it is to accent the mid frequency prescence control function which would be a noticeable mid boost/cut relative to the other frequencies either side.
I checked this for real and got 0.15V at the 5k presence pot for a 1V speaker out. This is a value of 20 x log 0.15 = -16.5dB.
When doing the FR test, I can only hear frequencies below about 15k anyway, even though those above this are boosted as seen in the plot.
All that remains is to take it to DBS tomorrow for a blast at high volume.
Really pleased with this amp fix as a project as I learned and re-capped a lot.