Yes the source resistor can be split. I see how Patrick did it with his versions. I will do that. Thanks!
From running simulations of EUVL's model, it appears that the output clips well before 30 watts into 8 ohms. I had to raise the rails to + / - 26V to obtain the voltage swing (43.7v p-p).
The FW J2 manual states 200W power draw. That would be 50W per device. At 26V, that would be 1.92A. Does 1.92A bias sound correct?
The FW J2 manual states 200W power draw. That would be 50W per device. At 26V, that would be 1.92A. Does 1.92A bias sound correct?
Point well taken. My first order assumption is that
Output Devices power consumption >> everything else
Everything else
1 Transformer losses
2 power resistors' consumption
3 power supply rectifier losses
4 ????
If everything else is < 10% of power consumption, then then the 1.9A bias guess is no more than 10% high.
If I am missing something, I will be happy to learn.
I noodled around with the model with high bias. Too much power into 4 ohms to match the User Manual specs.
My guess is that the power draw is worst case at 60 watts output.
So, subtracting 60 watts output from 200 watts and dividing by 4 and then dividing by 26 volt rails gives 1.35A bias. This assumes zero power losses in the transformer, rectifiers, source resistors, etc.
My guess is that the power draw is worst case at 60 watts output.
So, subtracting 60 watts output from 200 watts and dividing by 4 and then dividing by 26 volt rails gives 1.35A bias. This assumes zero power losses in the transformer, rectifiers, source resistors, etc.
I have re-worked EUVL's initial settings to achieve 30W at 8 ohms with nearly symmetric gain up to onset of clipping.
It appears to me that + / - 27V is required to achieve 30W at 8 ohms just before the onset of visible clipping.
Bias is probably still too high but much less than 2A.
It looks like gain is too high compared to the User Manual. Add more feedback.
Another idea that occurs to me is to replace the power JFET with a Tokin SIT. I would possibly cascode the input JFETs an additional 6 or more volts below the source of the SIT. This way the amp can remain DC coupled. It will cost an additional floating power supply. No problem for DIY.
It appears to me that + / - 27V is required to achieve 30W at 8 ohms just before the onset of visible clipping.
Bias is probably still too high but much less than 2A.
It looks like gain is too high compared to the User Manual. Add more feedback.
Another idea that occurs to me is to replace the power JFET with a Tokin SIT. I would possibly cascode the input JFETs an additional 6 or more volts below the source of the SIT. This way the amp can remain DC coupled. It will cost an additional floating power supply. No problem for DIY.
Some observations. At 1.8-ish amp bias, the 8 ohm power output is 30w and the 4 ohm power is 40w.
At 1.35A bias, 8 ohm power is 29w and 4 ohm power is 22w.
To get to 15w at 4 ohms, bias should be reduced below 1.35A.
When I build, I am going to stay above 1.35A bias. I have the IXYS current source lashed up and trying various resistors to get the bias and proportion of the center tap that I am looking for. The proportion is 1/5 of total resistance at the top and 4/5 at the bottom. This will be for a first-pass to get the output signal on a scope.
At 1.35A bias, 8 ohm power is 29w and 4 ohm power is 22w.
To get to 15w at 4 ohms, bias should be reduced below 1.35A.
When I build, I am going to stay above 1.35A bias. I have the IXYS current source lashed up and trying various resistors to get the bias and proportion of the center tap that I am looking for. The proportion is 1/5 of total resistance at the top and 4/5 at the bottom. This will be for a first-pass to get the output signal on a scope.
J2
Well people, I'm going to put the power to the first channel tomorrow. I could do it today, but I love the process and am in no real hurry. So tomorrow is the day, let's hope there's no smoke!
More resistors than needed, but they were just sitting there, so... 🙂 Also, you will will notice an ah, unorthodox rectifier I had some expensive halfwave fast soft laying around, and cobbled them together for a full wave.
Also, per Jeff, I did not sink the resistors to the main sink, but got some snap-on. He said they really didn't get hot enough to worry about them, but I still thought it prudent, when they cost about .20.
You may also note, one of the bleed resistors doesn't appear on the board... There was a senior moment and the wrong value was installed. In the process of take them out it got a little harry, so I mounted that one on the bottom of the board.
Well people, I'm going to put the power to the first channel tomorrow. I could do it today, but I love the process and am in no real hurry. So tomorrow is the day, let's hope there's no smoke!
More resistors than needed, but they were just sitting there, so... 🙂 Also, you will will notice an ah, unorthodox rectifier I had some expensive halfwave fast soft laying around, and cobbled them together for a full wave.
Also, per Jeff, I did not sink the resistors to the main sink, but got some snap-on. He said they really didn't get hot enough to worry about them, but I still thought it prudent, when they cost about .20.
You may also note, one of the bleed resistors doesn't appear on the board... There was a senior moment and the wrong value was installed. In the process of take them out it got a little harry, so I mounted that one on the bottom of the board.
Attachments
Last edited:
For the front-end of J2, two 2sk170 and two 2sj109 are used. 2sj109 is 2sj109BL. How about 2sk170? 2sk170GR, or 2sk170BL?
by default it's 2SK370GR, but i have implemented below advise:
"I suppose their combination is around 8mA.
You could take one of the K370 ( 4-5mA supposed) and one of your 7-8mA K170.
So you get an impression what the changes are.
Do not forget to readjust the offset!"
Thank you!
"readjust the offset!" means turning volume pot (variable register) until output become zero?
"readjust the offset!" means turning volume pot (variable register) until output become zero?
I have to say, that the J2 isn't the F5 little B. I have some 95+ speakers and this little amp is kicking my ***. So far, it's my fav Pass creation, no question at all. I love vibes and tube sound, however that is defined, but this little 25Watt effort is a amazing. I pushed it to the point where I thought the lower end would cry, and it never happened. 90db and it was rock solid playing RUSH and clearly, this is not optimum. Didn't matter, it performs amazing even out of it's intended element. Seriously impressed by this little 25W amp.
Hi all,
The 2sk170 better to be GR or BL range for the J2??
Any idea where to get the semisouth and j109 at decent price ?
I found some SJEP on a e-store at 100$/piece ��
Thanks
The 2sk170 better to be GR or BL range for the J2??
Any idea where to get the semisouth and j109 at decent price ?
I found some SJEP on a e-store at 100$/piece ��
Thanks
I think you're looking for 2SK170s that add up to 8 to 12mA. That will be possible with BL, but much easier with GR.
2SJ109s are as rare as hen's teeth. If you don't already have some and don't enjoy the quest for rare parts, use pairs of 2SJ74s (or LSJ74s). The latter can be got from the diyAudio store.
SemiSouths are a little easier to find, but seriously expensive. Here's a seller several others have used:
SemiSouth JFET Transistor SJEP120R100 High Performance Audio Amp New Matched x 2 | eBay
In all cases watch out for counterfeit parts.
2SJ109s are as rare as hen's teeth. If you don't already have some and don't enjoy the quest for rare parts, use pairs of 2SJ74s (or LSJ74s). The latter can be got from the diyAudio store.
SemiSouths are a little easier to find, but seriously expensive. Here's a seller several others have used:
SemiSouth JFET Transistor SJEP120R100 High Performance Audio Amp New Matched x 2 | eBay
In all cases watch out for counterfeit parts.
BL grade Jfets.
Cheap 109s? No such thing, and honestly, assume all every single one) on eBay to be counterfeit.
Cheap 109s? No such thing, and honestly, assume all every single one) on eBay to be counterfeit.
Two parallel GR's will dissipate more heat.
They have A grade LSK170 in the store, or will have shortly.
They have A grade LSK170 in the store, or will have shortly.