alexcp

Of course they would.

This is quite normal. As the heatsink temperature rises, the Vbe of Q14 goes down (by about 2 mV/°C), the voltage between the bases of NPN and PNP output transistors decreases, the quiescent current of the output transistors goes down, which reduces the dissipation, causing the temperature to go down. In other words, the Vbe multiplier adds a negative temperature feedback loop, which maintains not the quiescent current (which it has no way of measuring) but the temperature of Q14. Eventually the system finds an equilibrium somewhere, and the temperature at equilibrium depends, among other things, on the thermal resistance of the heatsink. In practical terms, you adjust the bias to your liking (e.g. by the lowest distortion, or by the desired heatsink temperature, or by the power dissipated by the output transistors) and be happy.

KSA-5 design is not particularly demanding on transistors, as lots of local feedback (aka degeneration) and the low impedances tend to even things out. Pretty much anything with the correct pinout (base in the middle, unless you want to carefully bend them) and with the rated Vce of 45V or higher should work fine. It might make sense to match the paralleled and differential-pair transistors, but even that is just gilding the lily.

The complexity of power traces looks excessive to me and suggests that an improvement could be made. If this was my board, I would try to shake it a bit, simplify, streamline, and make it more elegant. As an idea, you can try a different order of output transistors - instead of NPN-NPN-Thermal-PNP-PNP, something like NPN-PNP-NPN-PNP-Thermal, with power rails, the output trace and the ground running parallel to the edge of the board. This will bring in a different layout, which may (or may not) be helpful. Unless you plan to use some kind of off-board headphone protection, it would make sense to put the relay back. If you're not religious about the schematic, I'd suggest making the relay work for headphone protection, not just for turn-on delay. The amp seems to be reliable, but if you can avoid ruining some expensive headphones, why skip it? And of course - what would you expect - it would be great to make it possible to implement my mods (described above) on the PCB. It only takes two extra resistors and two small capacitors per channel, but the results are totally worthwhile.

That letter and the saturation current it signifies are not related to gain, either. One thing you're spot on is that in that particular place on KSA-5 clone schematic, the letter does not matter. In fact, JFETs can be replaced there with a pair of BJTs with a benefit.

This answer makes no sense, as hfe is the forward current gain of a BJT transistor. LSK389 is a pair of JFETs that have no appreciable gate current, so hfe is not applicable here. Instead, the letter specifies the drain saturation current group: The drain saturation current is the current that flows through JFET when you short gate to source and apply the specified voltage (here, 10V) between source and drain. The parameter has wide dispersion in the normal manufacturing process, so each manufactured JFET is measured and marked according to its saturation current. I have not seen LSK389 without a letter - are they genuine? According to the datasheet, every transistor should be marked with a letter:

Thank you! The final mod, that of the front end, is only slightly more complicated. Here is the updated schematic with both mods: The list of changes vs. the original schematic: Replace R1, R2, R6, R7 with 100 ohm resistors Replace R5 and R8 with 332 ohm resistors Replace R9, R10, R11, R12 with 274 ohm resistors Replace R16 and R17 with 22 ohm resistors Replace R19 with a 562 ohm resistor (reuse one of R5/R8) Replace R23 with a 1 Megohm resistor Replace R24 with a 68pF 50V NP0/C0G ceramic capacitor Replace R33, R34, R35 and R36 with 0.22 ohm 2W or 3W resistors Replace R37 and R38 with one 47 ohm resistor connected between bases of Q23/Q24 and Q25/26. Make sure to connect the new resistor correctly - see the photo below - or your output transistors are at risk. Replace R47 with a 3.92k resistor Remove C2 and C3 Add a 4.7nF film capacitor and a 47 ohm resistor, connected in series, between the collectors of Q2 and Q3. Add another 4.7nF film capacitor and a 47 ohm resistor, also connected in series, between the collectors of Q7 and Q8. Place the new parts on the underside of the board if you like. Repeat for the other channel. That's it! The list of parts required to modify two channels: 8x 0.22 ohm 2W or 3W resistors 4x 22 ohm resistors 6x 47 ohm resistors 8x 100 ohm resistors 8x 274 ohm resistors 4x 332 ohm resistors 2x 562 ohm resistors (not needed if you can reuse R5/R8) 2x 3.92k resistors 2x 1 Megohm resistors 2x 68pF 50V capacitors, NP0/C0G ceramic or mica 4x 4.7nF film capacitors Some 18 AWG single core insulated wire and 2x 10 ohm 2-3W resistors for the output RL network As resistors's names are not marked on the board, here is a photo showing what to replace with what. Note that the additional parts from step 12 above are not shown - they are under the PCB. After assembly, take the usual precautions before powering the amplifier up, as if it were a newly assembled board. Turn the bias adjustment trimpots all the way counterclockwise, connect a current limited +/-21V power supply with the current limit set at 0.5A per rail. With power on, check the output for a possible oscillation, then adjust the bias and re-check for oscillations. The bias level needs to be at about 20mA per transistor - with 0.22 ohm emitter resistors, it corresponds to 4mV between the test points, which should be easy to measure with a DVM. Higher bias levels are possible, but the distortion will be slightly higher. Let the amplifier warm up for 10-15 minutes and readjust the bias - it should go down as the output transistors warm up. As with any feedback amplifier, capacitive loads may affect stability. In my testing, the amplifier remained stable with capacitive loads of up to 100nF. Consider adding the usual RL network between the output of each channel and the load to ensure stability. Make 20-30 turns of 18 AWG single core insulated wire on a 1/2 inch (12mm) former - a Sharpie will work - to make an air core inductor like this: then connect a 10ohm 2-3W resistor in parallel to it.

I completed the full modification (output stage and front end) on both channels and listened to the amp briefly. The headphones were Grado GS1000 and Sennheiser HD595, the 8ohm speaker were B&W 602.5 floorstanders. The amplifier performed very well in each case and sounded immaculate. It is a HUGE upgrade over the original and over the output stage mod alone. I compared it against Musical Fidelity X-CANv8, and they performed equaly well. I took some measurements. Here it is delivering 1W and 5W with an 8ohm load: Here is the performance of the fully modified KSA-5 into a 32 ohm load: and into 100ohm:

Let me first modify the output stage. The mod affects only the performance with lower impedance loads, and even there it can take us only so far, but it is a start. The front end modification that increases the feedback loop gain will be posted separately. The changes are simple. Here is the schematic: R33-R36 are replaced with 0.22ohm 3W resistors, R19 is reduced to 470..560ohm to allow proper biasing, and R37-R38 are replaced by a single 47..51ohm resistor. The bias will need to be re-adjusted. Note that 50mA per transistor that resulted in 100mV between the test points in the original will now give 11mV - still easy to measure with a DVM. The performance improvements are as follows. With 1W and 5W into an 8ohm load: With 4Vpeak (2.8Vrms) into a 33ohm load the effect is much smaller: Distortion into 100ohm does not change appreciably, so I don't show it here.

To see what and how to improve, let's have a look at the original schematic. KSA-5 is designed along the lines of "moderate feedback", that is, it uses very little to no global feedback but lots of local feedback a.k.a degeneration. The pair of input JFET buffers (Q1, red box on the schematic above) run independently of each other and outside of the global feedback loop. With low loop gain, they see very different signal levels, so the differential stage downstream doesn't cancel their distortion. (BTW, because of this JFETs need not be matched. Also, the expensive and hard-to-find JFETs can be easily replaced here with BJTs.) Still, a JFET follower loaded by a current source has 100% degeneration and relatively low distortion, at least at low signal levels, so the buffers are not the biggest problem. The pair of differential stages (Q2+Q3, Q7+Q8, orange box) is heavily degenerated by 680ohm emitter resistors and produce R10/(R1+R2) = 2 = 6dB of gain. The pair of common emitter stages (Q12, Q13, purple box) is also heavily generated by 402ohm emitter resistors and, with the low load of R23 and R24, provides R23/R16 = 9 = 19dB of gain. Since the output stage (blue box) is a double emitter follower with approximately unity gain, the total open loop gain of KSA-5 is 2x9 = 18 = 25dB. The feedback divider (R45-R47) attenuates the output signal by a factor of 9 (19dB), which leaves 18/9 =2 (6dB) of global feedback. That is, the global feedback loop attenuates the distortion of the output stage by a small factor of 1+2 = 3. The output stage, meanwhile, is a large source of distortion. Although Krell claimed that KSA-5 runs in "pure Class A", in reality it can easily slide into Class AB. The output pairs run at only 50mA of quiescent current each and leave Class A (that is, one half of the output stage stops conducting current) when the output current reaches 200mA. The driver quads (Q15-Q22) also run in Class AB (R37 and R38 are connected to the output), which means they stop conducting at that point, too. With a 100ohm load, it would happen at 20V peak output voltage, so the amp never leaves Class A with such a load. However, with 32ohm, KSA-5 leaves Class A at 6.4V peak; with 8ohm, at 1.6V. Even within Class A region, the output stage is not very linear, especially with low impedance loads. It uses paralleled transistors with relatively large emitter resistors to ensure current sharing. The dark side of large emitter resistors is that they make the output impedance of the emitter follower large and nonlinear in the crossover region (see e.g. Douglas Self and his "wingspread" diagrams). Since the output impedance forms a voltage divider with the load, its nonlinearity makes the gain of the emitter follower nonlinear, adding crossover distortion and negating the benefit of the large bias current. Overall, KSA-5 has a nice and linear front end followed by a not-so-linear output stage, with little feedback to let the former help the latter stay linear. The game plan, then, is to improve the output stage and add more feedback.

With an 8 ohm load, Class A ends at 1.6V peak, so 1W performance is not so good anymore. Curiously, the Krell brochure mentions that the amplifier is good for 5W into 8ohm, but the owner's manual warns agains connecting it to any loudspeakers. I understand the 8ohm was meant for driving STAX via a transformer. As usual, at a higher output level the distortion percentage improves somewhat:

With a 32 ohm load, Class A extends to 6.4 volts peak, so at a lower level of 4Vpeak the performance is quite similar:

Let me start with a few measurements of the unmodified KSA-5. They seem to confirm the listening experience described above. With a 100 ohm load, KSA-5 never leaves Class A, and the performance is decent, with 0.02% THD and dominant 2nd and 3rd harmonics:

I built my KSA-5 clone back in 2013 (see my post dated August 9, 2013 in this thread on page 8) but was never quite satisfied with it. I made every effort to make it look good and perform well, and it was well within the original KSA-5's specs. Yet, although it worked well with my 32ohm Grado headphones, it could not compete, in my subjective opinion, with Musical Fidelity's X-CANv8. Connected to a pair of 8ohm speakers, the clone would become rather confused with anything but simplest music. Because of this, the amplifier fell into disuse and was gathering dust on my rack. Until this weekend. This weekend, I finally got around to make KSA-5 work for me. I kept the overall topology and the PCB, only changing some passive parts. The results are quite remarkable: The original KSA-5 was rated for 5W into 8ohm with THD<0.5%. Before modification, my clone gave 5W into 8ohm with THD @1kHz of 0.18%, well within the specs. The revised clone delivers 5W into 8ohm with 0.0015% THD, an improvement of more than two orders of magnitude. The original KSA-5 was advertised to deliver THD<0.03% into 100ohm load, although the brochure did not give the signal level for this performance. Assuming the same output voltage as for 5W into 8ohm, about 6.3Vrms @ 1kHz, my unmodified clone demonstrated THD of 0.02%, again within the specs. The revised clone drives a 100ohm load to the same level with 0.0017% THD, an improvement of more than an order of magnitude. The modified channel sounds very well, clear and transparent. I still need to complete the second channel and do some real listening. I will post the revised schematic with some explanations, as well as additional measurements, in this thread.

I matched the quad of Q23-Q26 by hFE, and separately matched the octet Q15-Q22. However, my main reason for matching was because I could. You cannot get perfect symmetry between NPN and PNP transistors no matter what. In my case, I could not match MJE15030 with MJE15031 from the batch I had, so I replaced them with D44H and D45H, respectively - these matched much better.

50C is normal for power semiconductors, you don’t need to make it cooler.

A CRC filter is something along these lines: In that diyAudio post, both capacitors were 4700uF and the resistor was 4ohm. Have a look e.g. here: http://www.learnabout-electronics.org/PSU/psu12.php for the theory behind it.

FIf you are working with the "new" power supply, the one with opamps, then each opamp compares the reference 12V voltage (supplied by D1 or D2) with the portion of the output voltage from the R8/R7 or R9/R10 voltage divider. If the voltages differ, the opamp drives the pass transistor so that the voltages become equal. With the 511/750 ohm divider, V(out) = 12V * (511+750) / 750 = approx. 21.2V. If you want a lower output voltage, you can reduce the value of the top resistor in each divider (R8 and R9). For example, 499ohm instead of 511ohm would give you 12V * (499+750) / 750 = approx. 20V, 475ohm would give you 19.6V, and so on. Paralleling 2kohm with the 511ohm gives you 1/(1/511+1/2000) = 407ohm and approx 18.5V. However, if you bias is too high, a better thing to do may be to carefully tweak resistors around V(be) multiplier (Q14) in the amp itself. Reducing R19 from 825ohm and/or increasing R18 from 221ohm would decrease bias. Be careful though - should R19 be missing, your bias current would be very high.

Thank you Kevin!

Hello, Another build in progress: The amp proper is built (see photos) and tested; the sound with Grado GS is wonderful! Thank you Kevin for sharing the design! No power supply yet; I use my desktop lab supply for now The caps are Samwha RD; perhaps not in the spirit of the thread, but I just had them The output transistors are D44H11 and D45H11 (I could not match MJFs, so I used what I could match; had to add some insulation though) Resistors are RN55/RN60 and Panasonic ERX3 No case yet, plan to use one from modushop/HIFI2000 Question: I want to use a high resistance volume control pot (50k or 100k). Would it be ok to increase the resistance between the gate of the input JFET and the ground so as not to load the pot too much? What would be a practical limit on this? 1Meg? Thank you!