Welcome! Need support, you got it. Or share you ideas and experiences.
210.245.182.196
Originally I wasn't very interested in adding a digital circuit to my Ella but Brian convinced me that he had made sure no digital noise would affect the sound so I changed my mind.
The first job was to change a resistor on each ABS board. It comes with a 3K3 which sets the bias to 40mA but with KT88's I wanted 50mA. A table of values included in the instructions says that means I'd need a 4K3. The problem is that it would then be fixed, but if I ever changed back to EL34 (or 6550) I'd have to swap the resistor again. So I decided to use one of the existing bias pots in parallel with the resistor and wanted a range of 0mA to 60mA, which the table said would need a resistance range from 0R to 5K6. With a 22K pot in parallel that meant using a 7K5 resistor, because (1/22) + (1/7.5) = (1/5.6)
Then I had to make some space for the boards, which meant moving the bias bridge rectifier, removing the additional bias capacitors I'd added (which probably aren't needed now due to the extra filtering in the ABS) and rerouting the main A/C supply wires.
After that it was easy to mount the boards on L-shaped brackets off the tube sockets. I would have preferred to have points to solder the wires to on the boards instead of the screw fittings but I suppose if you use a fixed resistor instead of a parallel pot then you want the board as easy to remove as possible so you can get to the resistor to change it. And it would be nice if the LEDs on the boards could be moved so they were chassis mounted but I suppose that would make the manufacture of the boards much harder.
In use the bias goes to the value set and stays there regardless of the amp temperature and it does keep the pairs of tubes nicely balanced. Because I had already changed the bias diode to a schottky rectifier the improvement in sound wasn't as big as it would have been for a standard Ella, but there was still an improvement in details and bass. On a standard Ella the improvement would be huge.
Photos and a layout diagram on my web page.
yesterday we received our first new-style Ella with the ABS installed. The ABS locks down the bias and it's nice to see whether cold/hot/in between, all tube pairs are 100% balanced. since it only takes one or 2 milliamps of imbalance to seriously impact on the performance of the transformer (push pull Transformers have no air gap and cannot tolerate DC current), we now enjoy the full potential of the amplifier at all times.a little more:
The fixed resistor on the ABS is intentional to make installation as simple and easy as possible, just as the screw terminals are, as well as the very concept.The concept:
Take an existing fixed bias amp, select the correct resistor for it, fit with minimal changes and forget about bias. Very result oriented. No mucking about.
For those that dislike screw terminals and absolutely want to solder, the screw terminals are easily unsoldered. However these connections have absolutely minimal impact on sonics (unlike most connectors in amplifiers and tons less than tube sockets) and most people can strip a wire and place it in a screw terminal.
Equally, it is possible to desolder the LED’s and replace each by external LED’s on long wires, but with already 8 or 9 wires to each PCB, adding another 8 would have been overkill for most DIY’ers, so simplicity meant putting the LED’s on the PCB. Anyway, you only have to worry if they stay red and in future the matching PSU board will take care to shut the amp down and light a “sumpting wrong” light in case of red-light, so, no worries.
If using a Pot & Resistor combo to adjust the Bias it is recommended to take the two holes on the PCB holding the resistor to attach the wires in the usual fashion and wire the resistor directly to the adjustable resistor terminals.
In addition it may be worthwhile considering a series resistor to apply a lower limit, as adjusting the Bias to 0mA usually has little value.
Eg. in Goneill’s case: use an adjustment range of 30 – 60mA, using a 2K4 (2.4k) resistor in series with the combination of the 22k adjustable resistor paralleled with a fixed 3K9 (3.9k), which allows an extremely precise adjustment of the current over the sensible usable range, with the lowest resistance 2.4KOhm (I = 29mA) to 5.7KOhm (I = 60mA).
For fancy stuff use a series resistor plus pot in parallel with suitable resistor as here:
R s R p
---^^^^^^----+---^^^^^^---+---
| ^ |
| / |
+---^^^/^^---+
| /
+---+ R adj
To calculate:Rs = R from table for minimum current desired
Rp//Radj = (Rmax – Rs) where Rmax from the table for maximum current desired
For the parallel combo we can solve a complex equation but it is probably easier using a generic on-line parallel resistor calculator like this:
http://www.electronics2000.co.uk/calc/calcrsp.php
So, for Goneill’s example:
I min = 30mA
I max = 60mAR min = 2370 - nearest easy standard 2400R/2K4/2.4k
R max = 5620Rp // R adj = 5620 -2400 = 3220
Put the 22000 Ohm (22k) into the R1 field for the 2 resistor parallel calculator
Enter the total value (3220)
Press calculate R2 (in our notation Rp)
Result 3.772 Kohm, nearest preferred value 3K9/3900Ohm/3.9k
Again, for simplicity all this was not put on the PCB, as there are so many tubes and possible ranges that it is hard to come up with a single setup that fits all and is flexible. The worst case could have been two adjustable resistors on the PCB (to set lower and upper limits or current without external R) added to the (optional) external pot.
For the more advanced DIY’er the needed changes are not difficult, for others the ability to have a simple resistor take of all might be godsend.
Finally, yes, it is all analog. One can make a computer in a digital way or an analog way.
Before Russia gave in and copied IBM’s digital computers (in the 1960, before IC’s were invented) they used to do cutting edge analog and hybrid (as in analog for speed, digital add on for final precision) computers, that had impressive performance.
The ABS uses basically analog computing circuits based on this sadly forgotten technique, using the “always just right” paradigm compared to the digital principle of “always around 1/2LSB wrong”… ;-)
regards
Engineering Resource Team
DIYHFS
Take control! DIY.
I see your points about ease of install, and having even more wires for the LEDs really would make it look like spaghetti! Besides, I'm getting quite used to seeing a blue glow from inside the amp. I suppose my only worry is that the screws might come loose after time with the continual heating and cooling, but then none of the other fittings that screw into the amp have (so far!) so perhaps I'm just worrying too much.
Interesting idea on arranging the resistors. So you'd have the 22K pot in parallel to one resistor and then that would lead into another resistor in series? You got me thinking about it and I ended up doing some charts. I hope I got the calculations right on these!
First, this is what I did, with a 22K pot in parallel to a 7K5 resistor.
The chart shows the setting of the pot across the x-axis and the total resistance as the y-axis. Assuming an EL34 would want between 35mA and 45mA and a KT88 would want 45mA to 55mA I've added red lines to show the required resistance ranges (35mA=2K7, 45mA=3K9, 55mA=4K7):
![]()
So for my KT88s I'm setting the pot between 8K and 12K5. But with a 3K9 in parallel to the 22K and a 2K4 in series you get this:
![]()
While it sounds a better idea to have more range, you actually end up with less adjustment in the required ranges! But the idea is right, and if instead you use 3K3 in parallel with the 22K and then 2K in series it becomes this:
![]()
Although this loses a bit of current (max is now about 57mA) this gives the best amount of adjustment range and also has the advantage that the ABS already provides a 3K3 so you just need an additional 2K.
Great to hear that the whole thing is analog. No wonder you were so certain there would be no digital noise! Analog computers, now there is a reminder from the past. At university part of my course required us to run feedback/feedforward simulations using analog computers, which of course the university no longer had. So we used some software called LEANS - Lehigh Analog Simulator. Basically it was a program to run on a digital computer that mimicked the calculations of an analog computer.
I remember the first analog computer I ever encountered. It was an autopilot computer. It was amazing how they made it overshoot with just the right rise and fall times. I was supposed to be fixing it, Instead I spent all day examining the circuit and playing with it on the test rig. Analog computers are works of electronic art.I'll be saving my pennies to get a set for my Ella signature level 3.
thanks Graham. I guess with all those surface mount active devices one could easily suspect one amongst them must be a digital Devil -- but the ABS is all analog, all the time.
It's encouraging to read that despite the modifications you have already made over the stock bias supply, the ABS still gave you some improvement in sound. since the ABS eliminates DC current through the transformer core, the improved bass response to noticed is to be expected. Probably the bias voltage we apply to the grid is one of the most overlooked factors in our quest to squeeze ultimate performance from our amplifiers.
thanks for the write-up, and your well drawn out diagrams.
Brian
Take control! DIY.