Hi.
A friend of mine is working on an SV811-3 amp using the mentioned 5kohm OPT´s. A couple of days ago he fired up the prototype and ran some tests which showed some serious shortcomings in the upper frequency range , -6dB at 20kHz together with a dip at 13kHz.
Further measurements told us that it is the transformer that is the problem, the signal is OK when measured on the plate of the output tube. At first he hardly believed what he saw (two sets of transformers were tested do eliminate the risk of a single manfacturing fault), but to confirm it he made a whole new prototype using entirely different tubes and PSU.
Same thing happened again.
BTW: He also checked the stuff he uses to do the measurements and that equipment is just fine.
What is happening here?
All other Hammond 162(X) SE transformers reach 20k without problems, so why not the 1628??
A friend of mine is working on an SV811-3 amp using the mentioned 5kohm OPT´s. A couple of days ago he fired up the prototype and ran some tests which showed some serious shortcomings in the upper frequency range , -6dB at 20kHz together with a dip at 13kHz.
Further measurements told us that it is the transformer that is the problem, the signal is OK when measured on the plate of the output tube. At first he hardly believed what he saw (two sets of transformers were tested do eliminate the risk of a single manfacturing fault), but to confirm it he made a whole new prototype using entirely different tubes and PSU.
Same thing happened again.
BTW: He also checked the stuff he uses to do the measurements and that equipment is just fine.
What is happening here?
All other Hammond 162(X) SE transformers reach 20k without problems, so why not the 1628??
Is the cathode adequately (measurably) decoupled at 20kHz? If it weren't, ra would rise, and that could cause the HF loss in the transformer.
Another thought: SE output transformers can change their HF response depending on which way round the primary is connected.
Finally, I'm assuming that the transformer was intended for this application. Transformers have a limited number of octaves of bandwidth, and matching them to the application is a case of sitting that bandwidth symmetrically (in octaves) about 1kHz. It might be worth checking that you don't have an excessively good LF response. If you have the ability to measure distortion, drive the amplifier to full power, lower frequency, and look for a rise in distortion. A well-matched SE transformer is usually good to 25Hz.
Another thought: SE output transformers can change their HF response depending on which way round the primary is connected.
Finally, I'm assuming that the transformer was intended for this application. Transformers have a limited number of octaves of bandwidth, and matching them to the application is a case of sitting that bandwidth symmetrically (in octaves) about 1kHz. It might be worth checking that you don't have an excessively good LF response. If you have the ability to measure distortion, drive the amplifier to full power, lower frequency, and look for a rise in distortion. A well-matched SE transformer is usually good to 25Hz.
The 811-3 circuit employs fixed bias so there is nothing in the cathode to bypass (ok, 10 ohms or something for current measurement).
Besides, my friend is very competent when it comes to tube amp designs, he has designed and built lots of really good sounding amps. I honestly don´t think that he has made any simple mistakes here. Everything has been double-checked.
The Hammond datasheet specifies what wire that should connect to what and the transformer was wired that way.
SV811-3 is mentioned in Hammonds list of suitable triodes for the 1628SE, so I can´t see anything wrong here. I thought for a while that the transformers were mislabeled, but that doesn´t make sense since everything works fine up to 12-13kHz.
Bass response was 1db down at 20Hz if my memory serves.
The weird thing here is that, as I mentioned, the transformers were tested in a completely different circuit (EL34 SE) and showed exactly the same behaviour.
Besides, my friend is very competent when it comes to tube amp designs, he has designed and built lots of really good sounding amps. I honestly don´t think that he has made any simple mistakes here. Everything has been double-checked.
The Hammond datasheet specifies what wire that should connect to what and the transformer was wired that way.
SV811-3 is mentioned in Hammonds list of suitable triodes for the 1628SE, so I can´t see anything wrong here. I thought for a while that the transformers were mislabeled, but that doesn´t make sense since everything works fine up to 12-13kHz.
Bass response was 1db down at 20Hz if my memory serves.
The weird thing here is that, as I mentioned, the transformers were tested in a completely different circuit (EL34 SE) and showed exactly the same behaviour.
Hi John,
Could you explain that? In my language?
Cheers,
Bas
Edit: I mean why is it fine on the plate and messed up after the transformer? Does that mean that the psu does not have a low impedance?
PS. Olimex quoted me 35$ including shipping for 6 of those boards. I am going for 6 of those...to do some debugging..
Well, thinking of it...
If the signal is correct before the transformer but not after it, doesn´t that indicate that something is wrong with the transformer? But of course, the PSU is in series with the transformer and if the PSU impedance rises...
Still I don´t want to think that Hammond makes a transformer that falls at 13k and specifies it to 20k.
If the signal is correct before the transformer but not after it, doesn´t that indicate that something is wrong with the transformer? But of course, the PSU is in series with the transformer and if the PSU impedance rises...
Still I don´t want to think that Hammond makes a transformer that falls at 13k and specifies it to 20k.
Someone at audioasylum has measured this frequency response:
http://world.std.com/~doyle/845seBode.html
Seems like it is the transformers after all.
http://world.std.com/~doyle/845seBode.html
Seems like it is the transformers after all.
Bas,
I'm afraid that Dutch and Afrikaans are out of my reach, so I'll try English
If the power supply was perfect i.e. zero ohm source impedance, then any signal voltage on the anode (measured relative to ground) should be identical to the signal voltage developed across the output transformer's primary.
Where they differ, the voltage difference must be developed across something. In this case I was speculating that it was the power supply that was the "something".
Power supplies often have a different source impedance to different frequencies. If the impedance rose with frequency, then we might expect such a result as Fuling's friend found.
Fuling,
This is not adding up for me. Even if the transformer was poor, I would expect to see the same dips on the primary as secondary. Perhaps someone who knows more about transformers than me could straighten us out on this.
Hi,
Not that I know more about OPTs but wouldn't it be easier to isolate the OPT, measure the freq. response at various power settings and loads and draw conclusions from that?
Usually OPTs show their true colours when measured close to their max power rating but a roll off at 13KHz seems really lousy by any standard, I reckon.
Too bad...I was looking forward to adding another Afrikaans recipe to my cookbook...
Cheers,
Not that I know more about OPTs but wouldn't it be easier to isolate the OPT, measure the freq. response at various power settings and loads and draw conclusions from that?
Usually OPTs show their true colours when measured close to their max power rating but a roll off at 13KHz seems really lousy by any standard, I reckon.
Too bad...I was looking forward to adding another Afrikaans recipe to my cookbook...
Cheers,
I've reproduced the same results. This time with a fixed bias
845 on the 1628SE:
Amplitude Response of 1628SE
Now what is interesting is that this problem DISAPPEARS when
I use the 1628SE as a plate choke for a simple parafeed
configuration! You can see the pink line (parafeed) is
continuous out to about 25kHz before response begins to
fall !!
This plot was produced with an 8 ohm resistive load... Now the
question is how does it look with a real speaker - i.e. a reactive
load.
Despite this, people seem to really like listening to this amp.
FYI my driver is an 8532 tube driving an interstage...
Here's the schematics:
845 stuff
-- Jim
845 on the 1628SE:
Amplitude Response of 1628SE
Now what is interesting is that this problem DISAPPEARS when
I use the 1628SE as a plate choke for a simple parafeed
configuration! You can see the pink line (parafeed) is
continuous out to about 25kHz before response begins to
fall !!
This plot was produced with an 8 ohm resistive load... Now the
question is how does it look with a real speaker - i.e. a reactive
load.
Despite this, people seem to really like listening to this amp.
FYI my driver is an 8532 tube driving an interstage...
Here's the schematics:
845 stuff
-- Jim
1628SE test
I have a 1628SE here, so I tested it tonight with a signal generator and scope. I used a 2000 Ohm resistor in series with the 50 Ohm output of the sig. gen. so as to simulate the SV811-3 Rp. I don't see any dip in the response as others have, just a smooth roll off on the high Freq. end, then a peak at around 30 KHz. Obviously the signal level (20 V p-p) is much less than in a real amplifier, and no DC current. I also found that varying the 2000 Ohm resistor changes the rolloff frequency as would be expected. Higher resistance giving a lower freq. The idle current in the output tube will vary the Rp of the tube, so would change the rolloff if not 2K Ohm as stated in the SV811-3 data sheet for 80 mA idle current. Could try varying the idle current bias setting on the amplifier to see if the freq. changes due to Rp changes.
Another possiblility comes to mind, maybe a mechanical resonance is present in the transformer, especially with a plastic spacer for an air gap. Could be resonantly compressing the spacer. Could try tightening the xfmr. bolts a little to see if that affects the freq. or try stacking some heavy books on the side of the xfmr. with the I laminations to dampen any resonance.
I have a 1628SE here, so I tested it tonight with a signal generator and scope. I used a 2000 Ohm resistor in series with the 50 Ohm output of the sig. gen. so as to simulate the SV811-3 Rp. I don't see any dip in the response as others have, just a smooth roll off on the high Freq. end, then a peak at around 30 KHz. Obviously the signal level (20 V p-p) is much less than in a real amplifier, and no DC current. I also found that varying the 2000 Ohm resistor changes the rolloff frequency as would be expected. Higher resistance giving a lower freq. The idle current in the output tube will vary the Rp of the tube, so would change the rolloff if not 2K Ohm as stated in the SV811-3 data sheet for 80 mA idle current. Could try varying the idle current bias setting on the amplifier to see if the freq. changes due to Rp changes.
Another possiblility comes to mind, maybe a mechanical resonance is present in the transformer, especially with a plastic spacer for an air gap. Could be resonantly compressing the spacer. Could try tightening the xfmr. bolts a little to see if that affects the freq. or try stacking some heavy books on the side of the xfmr. with the I laminations to dampen any resonance.
Smoking-amp, you need to load the secondary... Did you load
the secondary?
If you read my earlier post from tonite, I indicate how I observed
that the problem "goes away" when you unload the secondary,
i.e. as one might do if they decided to use the 1628SE as a
plate choke for parafeed ; A role that it seems to do well at
-- Jim
the secondary?
If you read my earlier post from tonite, I indicate how I observed
that the problem "goes away" when you unload the secondary,
i.e. as one might do if they decided to use the 1628SE as a
plate choke for parafeed ; A role that it seems to do well at
-- Jim
re-tested with 8 Ohm load
My earlier post was tested without any load, so I went and re-did the test with 8 Ohms on the 8 Ohm sec. Now I DO find a significant drop at 15.77 KHz. I then retried without the 2000 Ohm series resistor from the generator but with the 8 Ohm load and the same thing, drops out around 15.77 KHz. Next I re-did tests with a 4 Ohm load on the 4 Ohm sec. and now I hardly see any dropout, with or without the 2000 Ohm gen. resistor, very strange. I don't hear any sign of mechanical resonance in the xfmr. in the 8 ohm case, so mech. resonance seems ruled out. I can only surmise that the extra turns portion for the 8 ohm secondary must have a lot of leakage inductance or capacitance. I am going to retry the test again with the two 4 Ohm secondaries in series now to see what happens. Then with the total secondary series connected.
My earlier post was tested without any load, so I went and re-did the test with 8 Ohms on the 8 Ohm sec. Now I DO find a significant drop at 15.77 KHz. I then retried without the 2000 Ohm series resistor from the generator but with the 8 Ohm load and the same thing, drops out around 15.77 KHz. Next I re-did tests with a 4 Ohm load on the 4 Ohm sec. and now I hardly see any dropout, with or without the 2000 Ohm gen. resistor, very strange. I don't hear any sign of mechanical resonance in the xfmr. in the 8 ohm case, so mech. resonance seems ruled out. I can only surmise that the extra turns portion for the 8 ohm secondary must have a lot of leakage inductance or capacitance. I am going to retry the test again with the two 4 Ohm secondaries in series now to see what happens. Then with the total secondary series connected.
Something really weird is going on here.
My friend did a quite interrestning experiment yesterday. He replaced the 1628SE with a small OPT from an old radio. Even though that tiny OPT was pushed way beyound its limits the resulting HF response was alot better than with the Hammond unit.
jrdmedford mentions that 1628SE works good as a plate choke, doesn´t that mean that the main problem must be in the coupling between prim and sec windings?
My friend did a quite interrestning experiment yesterday. He replaced the 1628SE with a small OPT from an old radio. Even though that tiny OPT was pushed way beyound its limits the resulting HF response was alot better than with the Hammond unit.
jrdmedford mentions that 1628SE works good as a plate choke, doesn´t that mean that the main problem must be in the coupling between prim and sec windings?
more tests
On further testing of the 1628SE, I can say that the problem is definately in the 4 to 8 Ohm section of the secondary (Grn/Yel to Yel). The two 4 Ohm sections look good, the 4 Ohm part of the 8 Ohm winding (Blk/Yel to Grn/Yel) having just a small dropout. Unfortunately, one can only get 16 Ohms resultant by series connecting the two 4 Ohm sections.
I also have a 1640SE xfmr. so I checked it also under load, but it does not appear to have this problem with freq. dropout at all. The 8 Ohm winding specifically looks fine. One would think that these two transformers would be wound the same as regards the secondary, apparently not, bizarre.
I think someone needs to check with Hammond to see what is going on here. I am tempted to take the end covers off the 1628SE to see how it is set up.
On further testing of the 1628SE, I can say that the problem is definately in the 4 to 8 Ohm section of the secondary (Grn/Yel to Yel). The two 4 Ohm sections look good, the 4 Ohm part of the 8 Ohm winding (Blk/Yel to Grn/Yel) having just a small dropout. Unfortunately, one can only get 16 Ohms resultant by series connecting the two 4 Ohm sections.
I also have a 1640SE xfmr. so I checked it also under load, but it does not appear to have this problem with freq. dropout at all. The 8 Ohm winding specifically looks fine. One would think that these two transformers would be wound the same as regards the secondary, apparently not, bizarre.
I think someone needs to check with Hammond to see what is going on here. I am tempted to take the end covers off the 1628SE to see how it is set up.
It seems that we can agree that there is some major fault in the construction somewhere (in the 4 to 8 ohms winding, it seems).
One big question is why only the 5k xfmr?
Their 1,25 and 2,5k´s are fine and the 6,5k is at least alot better than the 5k even though it too had some dip at ~13k or something.
My friend has talked to the Swedish importer about this too see if he can find out anything.
One big question is why only the 5k xfmr?
Their 1,25 and 2,5k´s are fine and the 6,5k is at least alot better than the 5k even though it too had some dip at ~13k or something.
My friend has talked to the Swedish importer about this too see if he can find out anything.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.