Ah, progress!
I clipped in the 10k:600 transformers, with 1k resistors across their secondaries (Shoog's suggestion). I like it!
Right now I have 6N6P (gain of about 16x) with a 15k plate load (6x rp) into the primary of the 10k:600 transformer. I think it sounds pretty good this way.
The gain is quite low in this config. I figure the 16x gain from the 6N6P is being stepped down to about 4x. So theoretically, I need 750mV peak signal in to the preamp to get full power out from the power amp. Since there are insertion losses, I'll bet that's more like 1V in to full power. The result is that with most CDs, I can turn my volume control all the way up and the amp just barely clips on the loud parts. Maybe this is a good thing?
Unfortunately, that means there's not quite enough gain for the couple of SACDs I have that are recorded at a very low level, supposedly to take advantage of the increased dynamic range from SACD format. I have a copy of "Standard Coltrane" from Analogue Productions that was mastered at a really low level and needs to be cranked way up. The Sony SACD version of Kind of Blue is also mastered at a lower than usual level. (And I am not comparing these to typical 'loudness wars' CDs.)
I figure using a higher gain tube should do the trick. 6DJ8 is about double the gain of 6N6P (or 5687), so I guess that as a mu-follower will have to be it.
--
PS -- Speaking of 6DJ8 mu-follower... The output Z of the 6DJ8 mu-follower is less than 500R. That throws off the RC in the RCL calculations. I can only get the Q down to 0.7 with Zout = 500R, Cseries = 150uF into L = 18H. In the ultrapath connected parafeed arrangement, what does the voltage rating of the C from the bottom of the secondary to the lower tube's cathode need to be? That would be the capacitor labeled 2~20uf in this schematic:
My guess is the full B+ voltage, and a little higher for turn-on surge. If so, that's going to be a large electrolytic, 150uF 400VDC.
--
I clipped in the 10k:600 transformers, with 1k resistors across their secondaries (Shoog's suggestion). I like it!
Right now I have 6N6P (gain of about 16x) with a 15k plate load (6x rp) into the primary of the 10k:600 transformer. I think it sounds pretty good this way.
Output Z of the 6N6 is about 2200R, output coupling cap is 9.1uF, inductance of the OPT primary is 18H. According to http://sim.okawa-denshi.jp/en/RLChikeisan.htm, that series RC and parallel L yields F3 of 12.4Hz, Q of 0.64 (just about critically damped).
The gain is quite low in this config. I figure the 16x gain from the 6N6P is being stepped down to about 4x. So theoretically, I need 750mV peak signal in to the preamp to get full power out from the power amp. Since there are insertion losses, I'll bet that's more like 1V in to full power. The result is that with most CDs, I can turn my volume control all the way up and the amp just barely clips on the loud parts. Maybe this is a good thing?
Unfortunately, that means there's not quite enough gain for the couple of SACDs I have that are recorded at a very low level, supposedly to take advantage of the increased dynamic range from SACD format. I have a copy of "Standard Coltrane" from Analogue Productions that was mastered at a really low level and needs to be cranked way up. The Sony SACD version of Kind of Blue is also mastered at a lower than usual level. (And I am not comparing these to typical 'loudness wars' CDs.)
I figure using a higher gain tube should do the trick. 6DJ8 is about double the gain of 6N6P (or 5687), so I guess that as a mu-follower will have to be it.
--
PS -- Speaking of 6DJ8 mu-follower... The output Z of the 6DJ8 mu-follower is less than 500R. That throws off the RC in the RCL calculations. I can only get the Q down to 0.7 with Zout = 500R, Cseries = 150uF into L = 18H. In the ultrapath connected parafeed arrangement, what does the voltage rating of the C from the bottom of the secondary to the lower tube's cathode need to be? That would be the capacitor labeled 2~20uf in this schematic:
My guess is the full B+ voltage, and a little higher for turn-on surge. If so, that's going to be a large electrolytic, 150uF 400VDC.
--
Last edited:
Here it is a whole year later, and I have two competing ways of getting balanced signal to my power amp.
I never did use the 10k:600 transformers. But I did learn how to use LTspice, and noticed that using a stepdown transformer like that makes the driving tube work pretty hard, which of course increases distortion. I don't think that's my goal.
Instead, I purchased a new pair of Edcor XSM15K-15K transformers (15k:15k, or 1:1, center tapped primary and secondary). The first Edcor 15k:15k transformers were the WSM version, which is on a smaller core. The primary inductance is small on those, no more than 15H. The XSM15K-15K is much better, at least 50H. Swapping those in brought an immediate improvement in bass response (no surprise there). I was able to decrease the output caps in my 6N6P line stage to 2uF (down from 10uF). The 2uF caps in the preamp makes it noticeably more listenable. I guess smaller value caps are indeed better sounding, if you can use them.
I've also completed a variation on SY's ImPasse preamp, using a 6CG7 voltage amp stage to a 6DJ8 cathodyne (concertina) phase splitter/output stage. The power supply is very basic, just an old Thordarson power transformer, 5Y3GT rectifier tube, CLC (47uF--16H--100uF), then each channel has its own RC (1k--100uF). Heaters are AC, with +50V bias to stay within heater-to-cathode limits.
In LTspice, both look like they should work well, with reasonable distortion levels. Again according to LTspice, the 6N6P line stage I have runs at about double the distortion of the 6CG7. With the two of them built, I can swap them in and out and compare the sound.
Transformer coupling is nice. I'm hearing nice solid images, but there is a sort of 'generalization' of low level details. It's like the sound is a little smoothed out. Very listenable, though. Maybe the transformers need to 'break in'?
The balanced out preamp (with the 6DJ8 cathodyne phase splitter) is nice too. It's more 'hi-fi' sounding, crisper highs, the bass might be a little tighter, more little details are audible.
I like both. I don't know which is 'better.'
I've attached a schematic of the actively balanced preamp.
--
I never did use the 10k:600 transformers. But I did learn how to use LTspice, and noticed that using a stepdown transformer like that makes the driving tube work pretty hard, which of course increases distortion. I don't think that's my goal.
Instead, I purchased a new pair of Edcor XSM15K-15K transformers (15k:15k, or 1:1, center tapped primary and secondary). The first Edcor 15k:15k transformers were the WSM version, which is on a smaller core. The primary inductance is small on those, no more than 15H. The XSM15K-15K is much better, at least 50H. Swapping those in brought an immediate improvement in bass response (no surprise there). I was able to decrease the output caps in my 6N6P line stage to 2uF (down from 10uF). The 2uF caps in the preamp makes it noticeably more listenable. I guess smaller value caps are indeed better sounding, if you can use them.
I've also completed a variation on SY's ImPasse preamp, using a 6CG7 voltage amp stage to a 6DJ8 cathodyne (concertina) phase splitter/output stage. The power supply is very basic, just an old Thordarson power transformer, 5Y3GT rectifier tube, CLC (47uF--16H--100uF), then each channel has its own RC (1k--100uF). Heaters are AC, with +50V bias to stay within heater-to-cathode limits.
In LTspice, both look like they should work well, with reasonable distortion levels. Again according to LTspice, the 6N6P line stage I have runs at about double the distortion of the 6CG7. With the two of them built, I can swap them in and out and compare the sound.
Transformer coupling is nice. I'm hearing nice solid images, but there is a sort of 'generalization' of low level details. It's like the sound is a little smoothed out. Very listenable, though. Maybe the transformers need to 'break in'?
The balanced out preamp (with the 6DJ8 cathodyne phase splitter) is nice too. It's more 'hi-fi' sounding, crisper highs, the bass might be a little tighter, more little details are audible.
I like both. I don't know which is 'better.'
I've attached a schematic of the actively balanced preamp.
--
Attachments
Last edited:
I have not read this thread in its entirety so I apologise if this point has been made before.
I think it is important to be aware of the difference between a balanced output and a differential one.
A balanced output is simply one where the impedances in the hot and cold arms are identical. It is the impedance balancing that achieves the CMRR. To gain the benefit of the CMRR, the receiving end needs to be differential. It is not necessary for the transmitting end to be differential as long as it is impedance balanced. This means you can use a single ended source - normally called an unbalanced output. To make it impedance balanced all that is necessary is to insert an impedance equal to the output impedance in the ground leg of the single ended source. The single ended output becomes the hot connection and the connection to the added impedance becomes the cold connection.
Many of the examples shown in previous posts have differential outputs. As mentioned above, these are not necessary to achieve a balanced connection BUT if they are used their output impedances MUST be equal if the benefits of balanced operation are to be achieved. Far too many differential outputs do not have equal output impedance and often worsen the CMRR.
Cheers
Ian
I think it is important to be aware of the difference between a balanced output and a differential one.
A balanced output is simply one where the impedances in the hot and cold arms are identical. It is the impedance balancing that achieves the CMRR. To gain the benefit of the CMRR, the receiving end needs to be differential. It is not necessary for the transmitting end to be differential as long as it is impedance balanced. This means you can use a single ended source - normally called an unbalanced output. To make it impedance balanced all that is necessary is to insert an impedance equal to the output impedance in the ground leg of the single ended source. The single ended output becomes the hot connection and the connection to the added impedance becomes the cold connection.
Many of the examples shown in previous posts have differential outputs. As mentioned above, these are not necessary to achieve a balanced connection BUT if they are used their output impedances MUST be equal if the benefits of balanced operation are to be achieved. Far too many differential outputs do not have equal output impedance and often worsen the CMRR.
Cheers
Ian
If a cathodyne (concertina) stage is driving a differential stage, does that constitute a balanced output from the differential stage?
What's happening in this setup is that the preamp is the phase splitter (the cathodyne stage on its output), which is then plugged into a power amp with balanced differential inputs. The differential stage drives a push-pull, class A output stage.
--
AJT, I noticed your signature. Speaking of designs DIYers want to build...
This setup is basically a Williamson type circuit in two chassis. And guess what? It works pretty well!
Another seminal push-pull design was the Acrosound UL-II. That one has a long tailed pair input stage, push-pull driver stage, push-pull output stage. I'm thinking of building a line amp with a 6N6P or 6SN7 LTP, ccs in the tail, etc. Since I'd only use it with a tube amp (high impedance inputs), I can do that.
The other common topology is the Mullard 5-20 style. When I use my single-ended line stage into a 1:1 transformer/phase splitter, and then go to the power amp (with balanced inputs), I've basically got a Mullard topology with a 1:1 transformer doing the phase splitting, instead of the driver stage doing that.
I built a couple of Mullard-style amps and always thought they sounded kind of murky in comparison to the Acro UL-II style ("dual-diff"). LTspice suggests that open loop harmonic distortion is always higher with a Mullard-style topology (all other things being the same). My hope is that the 1:1 input transformer as phase splitter allows the differential driver to run with lower distortion (and higher gain).
What I haven't tried is a Williamson-style circuit. This is my first one, and I think I like it.
--
This setup is basically a Williamson type circuit in two chassis. And guess what? It works pretty well!
Another seminal push-pull design was the Acrosound UL-II. That one has a long tailed pair input stage, push-pull driver stage, push-pull output stage. I'm thinking of building a line amp with a 6N6P or 6SN7 LTP, ccs in the tail, etc. Since I'd only use it with a tube amp (high impedance inputs), I can do that.
The other common topology is the Mullard 5-20 style. When I use my single-ended line stage into a 1:1 transformer/phase splitter, and then go to the power amp (with balanced inputs), I've basically got a Mullard topology with a 1:1 transformer doing the phase splitting, instead of the driver stage doing that.
I built a couple of Mullard-style amps and always thought they sounded kind of murky in comparison to the Acro UL-II style ("dual-diff"). LTspice suggests that open loop harmonic distortion is always higher with a Mullard-style topology (all other things being the same). My hope is that the 1:1 input transformer as phase splitter allows the differential driver to run with lower distortion (and higher gain).
What I haven't tried is a Williamson-style circuit. This is my first one, and I think I like it.
--
The cathodyne provides a differential output. it will be impedance balanced if its output impedance are equal.
Doug Self has an excellent article on balancing here:
Balanced Line Technology
Cheers
ian
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.