DIY Audio Projects Forum
http://diyaudioprojects.com/Forum/

KT120 Oddblocks
http://diyaudioprojects.com/Forum/viewtopic.php?f=9&t=3089
Page 8 of 49

Author:  gofar99 [ 20 Apr 2011, 17:26 ]
Post subject:  Re: KT120 Oddblocks

Hi All, The bigger trannies are enroute :thumbsup: Edcor shipped them today. They will be here in a day or so. Soon it will be possible to see what these big tubes can do.

Good listening
Bruce

Author:  gofar99 [ 02 May 2011, 14:52 ]
Post subject:  Re: KT120 Oddblocks

Hi Everyone, Trannies installed.... even without treaking the amps are monsters. With zero time on the tubes the ouput is clean to 22 WRMS with 24 more on tap with a bit more THD than I like. I'll be fixing that. It should not be too hard. I will have these amps (other stuff too) at the Lone Star Audio Fest in Dallas May 13-15. Google it and if you can visit ... cool. 8-)

Good listening
Bruce

Author:  poty [ 03 May 2011, 05:54 ]
Post subject:  Re: KT120 Oddblocks

gofar99 wrote:
Hi Everyone, Trannies installed....

Have the CCS voltage rising disappeared?
gofar99 wrote:
the ouput is clean to 22 WRMS with 24 more on tap with a bit more THD than I like. I'll be fixing that. It should not be too hard.

46W total in the future? Sounds promising!
For me it seems the bias should be higher. The output tubes have finite amplification, so to achieve 150-160mA swing in anode you should apply around 25-28V swing voltage to the grid (in ultralinear stage).

Author:  gofar99 [ 03 May 2011, 12:20 ]
Post subject:  Re: KT120 Oddblocks

Hi Poty, The voltage on the CCS drops with the increased current. If you think about it this is expected. The voltage swing on from the driver is in the +/-50 range and is sufficient to push the tubes to max (they run class A1) however, I do expect to have to add a mod I use on the little amps to increase the cathode voltage slightly to keep the swing a little further away from the 4 volt minimum on the 317

Good listening
Bruce.

Author:  poty [ 03 May 2011, 15:23 ]
Post subject:  Re: KT120 Oddblocks

gofar99 wrote:
Hi Poty, The voltage on the CCS drops with the increased current. If you think about it this is expected.

My question was not about biasing, but about
gofar99 wrote:
The DC voltage across the LM317 increases with output when you start to push the tubes.

I think this was the primary reason for ordering the new transformers.
gofar99 wrote:
The voltage swing on from the driver is in the +/-50 range and is sufficient to push the tubes to max (they run class A1)

The mentioned amount of the driver's voltage swing is practically useless if the following hasn't changed:
gofar99 wrote:
... 300mA for a pair... The voltage at the cathodes is in the 37-42 volt range on the 120s

You cannot exceed the cathodes voltages with the driver's swing.
gofar99 wrote:
I do expect to have to add a mod I use on the little amps to increase the cathode voltage slightly to keep the swing a little further away from the 4 volt minimum on the 317

To achieve more headroom it is preferable to raise B+ instead. It is practically inevitable to do so anyway.

Author:  gofar99 [ 03 May 2011, 20:44 ]
Post subject:  Re: KT120 Oddblocks

Hi Poty, True, as far as it goes. A quirk with using a LM317 as a ccs in the cathodes is that the cathodes can not swing all the way to zero. Effectively this restricts how much drive the lower (grid to ground) tube can see as drive. In this case about 38 volts -4.5. When you get close to this value the 317 does a few funny things, like sense it is being starved. The waveforms go down the tubes (nice play on words) rather abruptly and distortion goes way up. The solution to get full drive sounds odd, but when you think of how a tube is biased (the difference between grid and cathode) it makes sense. Just add a bit of positive voltage to the grids (through a high impedance). The amount can not be too much as it adds to the voltage across the cathodes and thus increases the heat dissipated by the CCS. Just increasing the B+ will not have the same effect in this particular application. You could trade off current for B+ and use less of one and more of the other (and still stay within the ratings of the tubes) but then if you increase the B+, in order to maintain the new current level the voltage on the CCS has to go up. I find that the best window for operation seems to be in the 400-450 B+. The trade off in dissipation in the CCS is more favorable than at other settings.

The actual reason for the new transformers was that the older ones had a lot of resistance in the windings and were not rated for the amount of current I was using.

Hope this helps explain how it works.

Good listening
Bruce

Author:  poty [ 04 May 2011, 06:03 ]
Post subject:  Re: KT120 Oddblocks

OK, It seems I can't ask my questions clearly! :)
I'm wondering about:
gofar99 wrote:
I have run into a quirk on the KT120 version I have now up and running. The DC voltage across the LM317 increases with output when you start to push the tubes.
Has the problem gone?

-------
Then:
gofar99 wrote:
Just add a bit of positive voltage to the grids (through a high impedance). The amount can not be too much as it adds to the voltage across the cathodes and thus increases the heat dissipated by the CCS. Just increasing the B+ will not have the same effect in this particular application. You could trade off current for B+ and use less of one and more of the other (and still stay within the ratings of the tubes) but then if you increase the B+, in order to maintain the new current level the voltage on the CCS has to go up. I find that the best window for operation seems to be in the 400-450 B+. The trade off in dissipation in the CCS is more favorable than at other settings.

Sorry for the big quotation. I think in the message above you forgot about rising the voltage on the CCS in case of adding small positive voltage to the grids too. The bigger dissipation level in that case is caused just by the voltage rising.
Increasing B+ on the other hand could help to linearise the characteristics of the output tube as soon as it is moved to the area where bigger swing is allowed. 450V B+, according to my counting, is the minimal appropriate voltage for the tube in that implementation.

Author:  poty [ 13 May 2011, 05:34 ]
Post subject:  Re: KT120 Oddblocks

gofar99 wrote:
...Just add a bit of positive voltage to the grids (through a high impedance). The amount can not be too much as it adds to the voltage across the cathodes and thus increases the heat dissipated by the CCS.

Playing with the schematic I've come to another couple of thoughts.
If we use the current schematic of the KT88 output stage (with some tweaking of the resistors' values of course) we have different resistors' values in the halves' grids (to ground). Earlier you mentioned adding two 1-2 MOhm resistors to the grids from B+. As soon as the grids-to-ground resistors differs, the resulting addition of the voltage will differ too for each grid. First thought out of it - made the grid-to-ground resistors equal for both grids = the resistor's value for "primary" tube. In the real situation there is no signal current through the grid-to-ground resistor of the "slave" tube, so the value of the resistor is not important. Then we could use identical B+-to-grid resistors symmetrically.
The second thought is about the 25 Ohm pot. In the KT120 version there is as much as (25/2)*0.16 = 2V voltage dropping on the each halve of the pot. It is additionally to the around 4.5V safe margin for LM-enabled CCS and some insignificant voltage dropping on 1 Ohm resistors for measuring balance. The main point - the voltage on the 25 Ohm pot can't be compensated by rising the voltage on the grids, so it mainly wastes power. As soon as we adding the B+-to-grid resistors for rising the grids potential it is preferable to transfer the balancing pot here. Then we'll recover the 2V from cathode circuit, 4.5V from the CCS, have balancing pot in the circuit without signal currents and as a side effect could measure the balance directly on the cathodes instead of comparing voltages on each 1 Ohm resistor.
What do you think about this?

Author:  gofar99 [ 15 May 2011, 16:32 ]
Post subject:  Re: KT120 Oddblocks

Hi, I'm not sure we are in sync on questions and answers. There are two reasonable ways to balance the idle current in the tubes. Both have advantages and unfortunately negatives. The way most of the oddwatts work is to use a balance pot in the cathodes. Depending on the position of the wiper, you can effectively have a range of adjustment of around 2 volts between the cathodes. The current for the pair will of course be constant. The one ohm resistors only allow for ease of measurement of the current ( by voltage drop across them ). The alternative way to adjust the balance is to omit the 25 ohm pot and feed a relatively small positive voltage to each grid through a high impedance. This is done with a high value pot. This method can be used to also overcome the 4 volt problem area in the LM317. It does this by raising the cathodes up by that (or more) amount. The problems with this is first the CCS must dissipate more power, a second issue is it it now possible to actually cut off one or the other tube abruptly.

The issue you imply with the B+ is correct as far as it goes. I have found that the circuit works well over a wide range of B+. If the B+ is lowered, the current can be increased and vice versa. There are some limits as the transformer winding ratios do not change. But using the impedances given there is a wide sweet spot and the efficiency does not suffer greatly.

All this is good stuff...except the odd behavior I noted is related to the protective circuitry built into the LM317. It apparently relates to a predictive protection from over dissipation. The thing that initially threw me off track was that it will regain function above that level. A sort of transitional region. I have been conversing with them on exactly what is occurring and if there is a fix. I can envision several possibilities, but need confirmation. I have been on the road for the past week and will not be back to my shop for a little while yet and can't test the ideas, so please be patient. BTW, the simplest solution that has been provided so far is to have 2 parallel 160ma CCS in place of a single 320. The smart folks tell me it will work fine. Another alternative is to bootstrap the 317 with a power FET.

Good Listening
Bruce

Author:  poty [ 16 May 2011, 02:50 ]
Post subject:  Re: KT120 Oddblocks

As CCS I'm now breadboarding the attached schematic. Until now - works well with KT120. As you can see I'm using +/-12V now, but you can use anything with basic filtering as low as 5V (from a spare heating winding) or use $3-4 dedicated small transformer. The common level setting circuit can be shared between the two OPAs in the case. For one channel you can use 134 version instead of 2134.

Page 8 of 49 All times are UTC - 6 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group
http://www.phpbb.com/