The value of the cathode bypass caps in any amp, including the SSE are often a subject of much debate. In fact there was a thread in a forum on a speaker builder web site where it was stated that I didn't know what I was doing becaused I recommended 1500 uF caps in both positions when the math revealed that only 270 uF or so was needed.
The math will reveal that a certain capacitor value will cause a 3db roll off at a certain low frequency. I don't remember the exact numbers now but I believe the 270 uF value provided for a 3db point of 45 Hz. There will be some roll off with considerable PHASE SHIFT starting at 100 Hz with this value. Those who use the big Edcor or Hammond OPT's can see full power output down to the 20 to 30 Hz region if the capacitor isn't the limiting factor.
In an amplifier with NO negative feedback, or ONLY local feedback around the output stage (the CFB used in the SSE), you can generally increase the bypass and coupling caps without bounds to extend the lower frequency response. The OPT will become the limiting factor. This is how I want an amp set up so that there is only one frequency response determining element, the OPT.
You CAN NOT do this in any amp incorporating GNFB! Any time a cap or inductor (including the OPT) is operated near its 3db frequency it will cause a phase shift. If some of this signal is returned to the input section of the amp these phase shifts can add up to turn negative feedback into positive feedback. This can happen at either end of the audio range. Increasing the coupling / bypass caps in a GNFB amp can cause low frequency instability or motorboating.
Pure math and real world testing don't match up when we consider that electrolytic capacitors are far from ideal components. Aluminum is not a perfect conductor so all electrolytics have some resistance associated with them (known as ESR). Their guts are wound up in a spiral adding some inductance (ESL) to the equation (simplified explanations).
At the time the SSE was being designed I had access to a fancy HP component analyzer at my lab at work that could measure these parameters. I got a handful of each reasonably priced electrolytic that Digikey and Mouser had which would fit the requirements for the SSE and TSE amps. I analyzed them and picked the ones that offered the best real world measurements over the entire audio range. The sweet spot for the value that worked best was 1500 uF, which is why I chose that value.
In a general sense the larger the value of an electrolytic, the better it behaves as a bypass cap at the low frequencies. However as the value is increased leading to larger physical size the "imperfections" tend to increase. The ESR (effective series resistance) tends to drop with increasing size, but the ESL (effective series inductance) tends to increase because there is more material coiled up inside the cap. At some point the ESL will tend to cause a resonance (damped, but still a peak) in the ultrasonic, or upper audio range. In some cases the measured capacitance value starts to drop as well. Larger capacitors tend to have a higher dissipation factor too. This is where some of the audio energy is simply lost in the dielectric as heat.
In the capacitors that I analyzed the turning point was quite high, in the 4000 to 10000uF range. This was however nearly 10 years ago. Sadly our group at work no longer needed the component analyzer and it was scrapped to make room for new stuff. The lines of Panasonic capacitors I evaluated have all been discontinued in favor of newer products.
I have no idea how the new stuff behaves, but I would say that 2200 uF is OK if it is a quality cap. You can bypass it with a single film cap in the 1 to 10 uF range to smooth out any high frequency anomalies. I found that random connections of several types of caps in parallel in the "more is better" thought can produce some ugly resonant messes especially if low quality parts are used.
The math will reveal that a certain capacitor value will cause a 3db roll off at a certain low frequency. I don't remember the exact numbers now but I believe the 270 uF value provided for a 3db point of 45 Hz. There will be some roll off with considerable PHASE SHIFT starting at 100 Hz with this value. Those who use the big Edcor or Hammond OPT's can see full power output down to the 20 to 30 Hz region if the capacitor isn't the limiting factor.
In an amplifier with NO negative feedback, or ONLY local feedback around the output stage (the CFB used in the SSE), you can generally increase the bypass and coupling caps without bounds to extend the lower frequency response. The OPT will become the limiting factor. This is how I want an amp set up so that there is only one frequency response determining element, the OPT.
You CAN NOT do this in any amp incorporating GNFB! Any time a cap or inductor (including the OPT) is operated near its 3db frequency it will cause a phase shift. If some of this signal is returned to the input section of the amp these phase shifts can add up to turn negative feedback into positive feedback. This can happen at either end of the audio range. Increasing the coupling / bypass caps in a GNFB amp can cause low frequency instability or motorboating.
Pure math and real world testing don't match up when we consider that electrolytic capacitors are far from ideal components. Aluminum is not a perfect conductor so all electrolytics have some resistance associated with them (known as ESR). Their guts are wound up in a spiral adding some inductance (ESL) to the equation (simplified explanations).
At the time the SSE was being designed I had access to a fancy HP component analyzer at my lab at work that could measure these parameters. I got a handful of each reasonably priced electrolytic that Digikey and Mouser had which would fit the requirements for the SSE and TSE amps. I analyzed them and picked the ones that offered the best real world measurements over the entire audio range. The sweet spot for the value that worked best was 1500 uF, which is why I chose that value.
In a general sense the larger the value of an electrolytic, the better it behaves as a bypass cap at the low frequencies. However as the value is increased leading to larger physical size the "imperfections" tend to increase. The ESR (effective series resistance) tends to drop with increasing size, but the ESL (effective series inductance) tends to increase because there is more material coiled up inside the cap. At some point the ESL will tend to cause a resonance (damped, but still a peak) in the ultrasonic, or upper audio range. In some cases the measured capacitance value starts to drop as well. Larger capacitors tend to have a higher dissipation factor too. This is where some of the audio energy is simply lost in the dielectric as heat.
In the capacitors that I analyzed the turning point was quite high, in the 4000 to 10000uF range. This was however nearly 10 years ago. Sadly our group at work no longer needed the component analyzer and it was scrapped to make room for new stuff. The lines of Panasonic capacitors I evaluated have all been discontinued in favor of newer products.
I have no idea how the new stuff behaves, but I would say that 2200 uF is OK if it is a quality cap. You can bypass it with a single film cap in the 1 to 10 uF range to smooth out any high frequency anomalies. I found that random connections of several types of caps in parallel in the "more is better" thought can produce some ugly resonant messes especially if low quality parts are used.
I never noticed these before. They seem to be sold like many "audio" things. Look at the data sheets and you see several nice words, some that make no sense, but no real data. Not an ESR or ESL spec to be found. They may be good parts, and once I get caught up on life, I might get a few ans test them against some more "normal" parts.
I didn't check the mechanicals to see if they fit the board, but if I had to call it I would choose the larger value.
I have bypassed the power tube's cathode cap with a .1 uF film cap:
Vishay R. Film Caps
I think it made a significant improvement. I think your suggestion for a cap in the 1 uF range is probably better. I'm thinking of trying a couple of these, they are a lot smaller than similar 600+V rated film caps.
ClarityCap ESA Series Film Capacitor
One question on the driver tube, does it really need a bypass cap at all? Would it be ok to remove it completely?
Also, I have heard some suggestion for putting a small value resistor before the bypass caps and was wondering if you have tried it?
Removing the bypass cap completely causes a significant reduction in gain and doesn't help the sound much, if at all. There is no harm in trying it, your results may be different. Adding a resistor in series with the cap can smooth out any ripple caused by imperfections in the cap. As the resistor value is increased, the gain drops, and the "smoothing" effect increases. I think I had 33 ohms in my amp for a while.
One of my amps has no cap, and no resistor......it uses an LED. LED bias works good on the 12AT7, but it takes a bit of experimenting to find a tube and 2 LED's (one for each channel) that play nice together. The other amp is built exactly according to the assembly manual.....6 years ago.
One of my amps has no cap, and no resistor......it uses an LED. LED bias works good on the 12AT7, but it takes a bit of experimenting to find a tube and 2 LED's (one for each channel) that play nice together. The other amp is built exactly according to the assembly manual.....6 years ago.
Interesting, I have about 6-9 dB too much gain so I may give it a shot. Even if it makes no difference the reduction in gain would be welcome.
I am using a 12BH7 driver tube (with a lot more CCS current vs 12AT7) because I built the SSE as a power amp, I had a nice preamp at the time, and just built an Aikido with 20 dB gain, when 12 dB would have been more ideal.
Do you think LED bias is an improvement, I have heard conflicting opinions on it? It does sound like it would require some experimentation, and possibly for no gain...
I am using a 12BH7 driver tube (with a lot more CCS current vs 12AT7) because I built the SSE as a power amp, I had a nice preamp at the time, and just built an Aikido with 20 dB gain, when 12 dB would have been more ideal.
Do you think LED bias is an improvement, I have heard conflicting opinions on it? It does sound like it would require some experimentation, and possibly for no gain...
Is it a must that C12, C22 be rated for 105 degrees C? I found the following, but rated for 85 C:
2222 021 18152 - VISHAY BC COMPONENTS - CAPACITOR ALUM ELEC 1500UF, 63V | Newark
105 C rated ones are non-stock items.
Thanks in advance.
2222 021 18152 - VISHAY BC COMPONENTS - CAPACITOR ALUM ELEC 1500UF, 63V | Newark
105 C rated ones are non-stock items.
Thanks in advance.
MCM Electronics: Home and Pro Audio/Video, Security and Test Equipment
They have 47uf and 120uf 105 degree caps in stock. I have been using them for a little while. They ship fast and are priced ok. Sound good to me. Can't really tell the difference between them and a name brand.
They have 47uf and 120uf 105 degree caps in stock. I have been using them for a little while. They ship fast and are priced ok. Sound good to me. Can't really tell the difference between them and a name brand.
Here's some good caps for cathode bypass! 
FFLI6B1607K-- - AVX - CAPACITOR PP FILM PP, 1600UF, | Newark
FFLI6B1607K-- - AVX - CAPACITOR PP FILM PP, 1600UF, | Newark
cathode bypass cap for 12at7
I notice here and elsewhere there is much conversation about the size, type and bypassing of the capacitor to use to bypass the cathode resistor of the output tube.
How about the input tube? It would seem to have a similar effect. Maybe more.
Anyone have any experience with this?
thanks.
I notice here and elsewhere there is much conversation about the size, type and bypassing of the capacitor to use to bypass the cathode resistor of the output tube.
How about the input tube? It would seem to have a similar effect. Maybe more.
Anyone have any experience with this?
thanks.
The bypass cap on the input tube performs the same function as the cap on the output tube, but sees far less current, so it is not as critical. I would still use a good quality cap because it is in the signal path.
A 1000 uF cap is OK, as is a 4700 uF (tried it because I had some) but I wouldn't stray too far outside this range. Too small of a cap will affect bass response, and too big could cause a warm up bias shift especially with a big coupling cap......I had strange results with 10,000 uF in the input tube bypass, 2200 in the output tube bypass, and 1 uF coupling caps. I tried them because I had them, but got some weird woofer motion that took a second or two to settle out when cranking some bass heavy music.
A 1000 uF cap is OK, as is a 4700 uF (tried it because I had some) but I wouldn't stray too far outside this range. Too small of a cap will affect bass response, and too big could cause a warm up bias shift especially with a big coupling cap......I had strange results with 10,000 uF in the input tube bypass, 2200 in the output tube bypass, and 1 uF coupling caps. I tried them because I had them, but got some weird woofer motion that took a second or two to settle out when cranking some bass heavy music.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.