Please consider donating to help us continue to serve you.
Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Chapter 2: References and Corrections
This chapter is lacking some of the key links to resources because we are unable to attach files to the pages right now. These resources will be loaded as soon as this issue is fixed.
Janís web page for the Krell Klone : http://www.delta-audio.com/Krell-Clone.htm (doesn't seem to be online anymore)
[Corrections to the Board (pdf)]
[Parts List (pdf)]
[Parts List/BOM Post #4237]
Important: For those using the Pinkmouse PCBs, some transistors used in the Pinkmouse PCB have a
different pin-out than those used in Jan's boards. Use the devices listed in Pinkmouse BOM.
- To build this amplifier without the voltage-current limiting protection circuitry (like the original), omit the following
I'm not sure this is correct, can someone verify this and then delete this line?)
Issues regarding R124. The value of R124 is not shown on the schematic. R124 should be 4.7k, as seen in original schematics and verified through independant calcs and measurements
C105 and C106
The schematic lists the values of C105 and C106 as 390 pF. Good results were obtained with 47 pF.
See this link: [Post 2440]
See this link: [Post 1019] Suggests using 33, 47, or 68 pF. Not 330 pF. And not 22 and 33 pF (post 1020).
It appears that there might be a slight difference of opinion on this.
The board has options for normal (R126) and low bias (R144). It is not shown is how to switch between these or how they are connected.
My guess-- A switch can be connected to short/open circuit the low/high bias pads, so, when the switch is flipped, the pads are shorted making R126 parallel to R144. This gives a lower resistance and increases bias. As such the board is mislabeled. This setup will have R126 as stand-alone for low bias, and the combination of R144 and R126 in parallel is high bias. Thus, first set R126 for low bias, and leave it alone; then flip to high bias and set R144 for the desired high bias current. Because the low bias resistor is 5K, the parallel resistor I chose for R144 is 25K, so that there is a wider range of tuning rather than having 5K in parallel with 5K (which gives a max setting of only 2.5K). This setup is shown in [Post #3532]. It was suggested less than 5K would be ok to put in parallel but 25K has been tested and works.
See also [Post #2530].
The original KSA-50 (and KSA100) were both true dual mono amplifiers all the way back to the line cord.
In the original Krell it was ~37.5 VDC, bias ~1.9A per channel, ~60w/8ohm class A, ~75w/8ohm class AB. People have built these unmodified with 39VDC
Krell used a 400va transformer per channel, ie two per stereo chassis. In the course of production they used both toroids and EI transformers so no big deal there.
A normal KSA50 channel has ~37v rails. This requires 2 secondaries at ~26v AC, or 52v center tapped. A minimal transformer for two channels would be about 400va, you'd want more for driving low impedance loads.
Each transformer was rectified then smoothed by a pair of 40000uF caps, so 2 x 40000 caps per channel, 4 per stereo chassis. A fine alternative: the 68000/50v 'computer grade' parts that seem quite readily and inexpensively
available, check with Steve at apexjr.
The design specifies .5 watt resistors. It has been stated that most resistors can be 1/4 watt (from [Post#2626]):
Actually my KSA-50s all have 1/4 watt resistors except for the ones in series with the zeners, the ones at the pre-drivers, and the drivers emitter resistors themselves. Those 4 are are 1/2 watt and the drivers get 2 watt. I get my
Metal films surplus for 5 cents each so 1/4 watt will suffuce at that price. Its also ok at 37 volt rails according to Stuart Easson.
The schematic indicates .5 ohm emitter resistors, RE1, RE2, RE3. These should be rated for at least 5w, in the event very low impedance loads are driven a larger fraction of the considerable output power will be dissipated in
Other values may be used and affect the amount of class A bias and ultimate power delivery.
Actually the 'normal' range of values of the emitter resistors (0.22-1ohm) doesnt really affect the 8ohm power much. For >2ohm loads I'd go with 0.5ohm or greater for better bias stability. If driving <2ohm loads to full power the resistors will need to be seriously high power, and here lower values will minimize losses...
With the original power supply of 37.5 VDC and .5 ohm resistors, the result will be ~55 watts of Class A power, and ~75 watts of class AB power into 8 ohms.
Plenty of heatsinking should be provided for this amp. Since this is a true class A design its going to dissipate right around 150 watts of heat per channel when biased for 50 watts class A power level. Air tunnels are reccomended for this and can be easily accomplished by placing two flat back sinks face to face and installing a pair of 4" fans below them. Arrange the fans so they blow up and out the top of the tunnel. The fans can be run at half speed or less but don't allow sink temperature to get over 60 C. or device reliability may be short. A temperature servo could also be an efffective addition keeping fans at just enough speed to keep the sinks at about 55 C. or lower.
The use of a 60 deg.c to 65 deg.c Clixon thermostatic switch mounted to each heat sink and wired back to the transformer primary is recomended as a safety precaution to shut down the amp and to save the output devices in
case of excessively high temperature or should a cooling fan fail or stall. These switches are inexpensive insurance, or be sure your home owners policy is up to date! As an alternative each Clixon Switch could be wired to drastically lower the bias of the overheating channel when it is activated by too high of temperature.
A graph of temperatures of such a fan cooled heatsink over 3 hours of elapsed time is posted at [Post #4375]. The 4 graphs are are for a fan cooled output stage (2 channels), a convection cooled driver stage (1 channel), and a convection cooled voltage regulator (for driving the fans, softstart etc...).
Suitable convection heatsinks will need to be extremely large. See [Post #4469] for how to compute the thermal resistance of a heatsink that can dissipate the heat sufficiently. User 'geezer1944' in [Post #4475] has provided a [link] to a website with additional heat sink calcualtions.
Mounting the driver devices and the bias sense device an inch or so apart on the main sink for each channel will aid in keeping the driver devices at a reasonable operating temperature. In fact if done this way they should not exceed the operating temp. of the amp. This method also provides excellent thermal tracking. This was one of the main reasons for Pink Mouse's board re-design... to easily allow booth drivers and the bias device to be mounted on the main heat sink.
However, it was stated in [Post #2448] that the driver Transistors need not be on the output heatsink for thermal
tracking, but the bias transistor should be mounted in some way to track the output device heat:
The Vbe multiplier, q111 has to track the temp of the outputs. The drivers are not required to track any particular temp, they just have to be kept at a safe temp.
As q111 gets hotter the bias voltage will decrease in proportion to the temp.
As the drivers and outputs get hot, the bias current increases in proportion to the temp. In a perfect world the reduction in voltage keeps the bias current pretty much the same...
The drivers dissipate power at a fairly constant fraction of the output stage proper (~1/40). So if you choose a separate sink for them and attach q111 to it, I think you achieve a good approximation of the necessary thermal
tracking, but the size of sink you choose needs to keep the temp of the drivers and q111 in the same ballpark as your outputs.
Don't know what thickness wire to use?
Wire thickness can be computed using the link provided in [Post #2677]. Here is the [link].
To calculate at what voltage the input will cause the output to clip, see [Post #2831], which states:
I had a problem with a ground loop at the inputs. Whenever both inputs were connected there was a 60 cycle hum present. I tried many different things until pooge posted this in [Post #3635]. I followed those instructions and now the amp is dead quiet.
|Search this Page|
|New To Site?||Need Help?|