I wonder how many people use tube preamps with high output impedance driving low input impedance ss power amps and are quite happy with the resulting HF response. By that I mean they don't feel that that the HF is rolled off or 'soft'.
I've been playing around with a preamp with an output impedance of around 8 K ohms. On my sim it shows the following performance with two different power amps:
amp1 : 77pF ( interconnect cable ) + 150 pF (amp input )= 225 pF total capacitance
Rin = 10 K ohm
20 Khz response = -0.07 dB / -7.2 deg
amp2 : 77pF ( interconnect cable ) + 275 pF (amp input )= 352 pF total capacitance
Rin = 20 K ohm.
20 Khz response = -0.271 dB / -14.2 deg
So it appears that if we keep the interconnect and other capacitances low , the high output impedance should be OK.
In a majority of cases the ss amp input impedance will be typically around 10 K ohms or greater. If it's DIY this is easy with most circuits.
It appears that we might not need a cathode follower stage in many cases. Distortion figures might vary with the different loading on the input stage but I wonder if that would make a significant audible difference (?).
Cheers.
I've been playing around with a preamp with an output impedance of around 8 K ohms. On my sim it shows the following performance with two different power amps:
amp1 : 77pF ( interconnect cable ) + 150 pF (amp input )= 225 pF total capacitance
Rin = 10 K ohm
20 Khz response = -0.07 dB / -7.2 deg
amp2 : 77pF ( interconnect cable ) + 275 pF (amp input )= 352 pF total capacitance
Rin = 20 K ohm.
20 Khz response = -0.271 dB / -14.2 deg
So it appears that if we keep the interconnect and other capacitances low , the high output impedance should be OK.
In a majority of cases the ss amp input impedance will be typically around 10 K ohms or greater. If it's DIY this is easy with most circuits.
It appears that we might not need a cathode follower stage in many cases. Distortion figures might vary with the different loading on the input stage but I wonder if that would make a significant audible difference (?).
Cheers.
it's not the input resistance that is the HF problem.
It's the RF filter components.
We usually aim for the RC of the low pass filter to be between 0.3us and 1.5us.
Your 8k source impedance restricts the cable and amplifier capacitance to between 37pF and 190pF.
Cable alone causes excessive RF filtering. The added capacitance inside the power amp pulls the filtering effect inside the audio band.
Rs=8k and Cload >=300pF will be noticed by the majority of listeners.
It's the RF filter components.
We usually aim for the RC of the low pass filter to be between 0.3us and 1.5us.
Your 8k source impedance restricts the cable and amplifier capacitance to between 37pF and 190pF.
Cable alone causes excessive RF filtering. The added capacitance inside the power amp pulls the filtering effect inside the audio band.
Rs=8k and Cload >=300pF will be noticed by the majority of listeners.
Moreover, depending on the preamp design, the lower loading could significantly reduce gain and increase distortion. For example, an open loop amplifier with output taken off the plate will probably have that plate load halved by connecting it to the SS amp. That means3-6dB or gain loss, and probably double the distortion.
"........significantly reduce gain and increase distortion........."
Yes I was aware of that. I was considering a preamp which can give enough voltage to drive the power amp with a 1 volt input signal and have acceptable levels of distortion. Using a CCS on the anode and a low external impedance ( load ) , it might be similar to using a resistive load on the anode. At least the sim indicates that.
Last night I completed the test board for a single tube preamp . Will try to test it today and see how it fares.
On many power amps I've seen the input RF filter cap values vary from about 33pF to 470pF. This would also depend on the over all input impedance and the series input resistor.
I just wanted to find out if we have all these parameters under control (or known ) and we use a high Zout preamp , would it sound good enough assuming we take care of the interconnect capacitance. I will have to check out a Cat5e type DIY IC also. Does anyone know what capacitance that has per meter ?
By the way I measured the capacitance of an average RCA socket as 5 pF. Will have to check that on some expensive RCA connectors that I have.
This preamp uses a CCS , unbypassed cathode resistor and has an ac heater supply. Simmed Zout is just about 8 Kohms. I'll be trying out a resistor loaded anode and a CCS loaded anode. I provided space on the pcb to take either load.
The CCS uses a MJE350 and BC556B . The power supply is RC filtered with a capacitance multiplier following that. Anode load is further filtered with a typical split resistor and capacitor to ground.
I have a few commercial power amps and some DIY power amps to try it on.
I can change the RF filtering on the commercial amps if required . I'll decide what to do after checking out the response with the preamp attached.
Will be doing some listening tests this weekend with a couple of friends and a few bottles of wine ( DIY too !).
Cheers.
Yes I was aware of that. I was considering a preamp which can give enough voltage to drive the power amp with a 1 volt input signal and have acceptable levels of distortion. Using a CCS on the anode and a low external impedance ( load ) , it might be similar to using a resistive load on the anode. At least the sim indicates that.
Last night I completed the test board for a single tube preamp . Will try to test it today and see how it fares.
On many power amps I've seen the input RF filter cap values vary from about 33pF to 470pF. This would also depend on the over all input impedance and the series input resistor.
I just wanted to find out if we have all these parameters under control (or known ) and we use a high Zout preamp , would it sound good enough assuming we take care of the interconnect capacitance. I will have to check out a Cat5e type DIY IC also. Does anyone know what capacitance that has per meter ?
By the way I measured the capacitance of an average RCA socket as 5 pF. Will have to check that on some expensive RCA connectors that I have.
This preamp uses a CCS , unbypassed cathode resistor and has an ac heater supply. Simmed Zout is just about 8 Kohms. I'll be trying out a resistor loaded anode and a CCS loaded anode. I provided space on the pcb to take either load.
The CCS uses a MJE350 and BC556B . The power supply is RC filtered with a capacitance multiplier following that. Anode load is further filtered with a typical split resistor and capacitor to ground.
I have a few commercial power amps and some DIY power amps to try it on.
I can change the RF filtering on the commercial amps if required . I'll decide what to do after checking out the response with the preamp attached.
Will be doing some listening tests this weekend with a couple of friends and a few bottles of wine ( DIY too !).
Cheers.
Last edited:
I recently installed ferrite beads and 1nF ceramic caps on my SSamp's inputs (the local radio station was getting quite annoying!). Doesn't sound any different. Preamp has 12AU7 CF outputs, Zo around 600 ohms let's say.
The amp itself has 10kohm "base leak" resistors.
I can't imagine using a preamp where the output impedance is so high (like 12AX7 high). HF response doesn't really matter, I would hate that simply because gain would vary so much, let alone distortion.
Stuff like 12B4 plate is almost 12AU7 cathode scale and no big deal; 6DJ8 follower would be quite excellent, as well as other high-Gm types.
Tim
The amp itself has 10kohm "base leak" resistors.
I can't imagine using a preamp where the output impedance is so high (like 12AX7 high). HF response doesn't really matter, I would hate that simply because gain would vary so much, let alone distortion.
Stuff like 12B4 plate is almost 12AU7 cathode scale and no big deal; 6DJ8 follower would be quite excellent, as well as other high-Gm types.
Tim
Quick preamp test.
I tested the tube preamp in a hurry tonight. Zout is high and appears to be a bit higher than 8K ohms. Response with a 8.7K ohm load and 320pF capacitance is -0.6dB at 20Khz ( measured). Distortion with a CCS is 0.03% at 2 V rms ! With a resistive load at the anode the distortion is much higher. With a CCS I see practically only the second harmonic the higher orders being very low . With a resistive anode load the third fourth and fifth are also significantly present though they keep decreasing in amplitude. So CCS is the way to go.
I rigged it up with one power amp and I must say that it sounds very nice. I can't play it loud as it's midnight here. Tomorrow or Sunday will be the day for regular listening sessions. I don't hear any loss of HF and it doesn't sound 'bright'. Voice sounds very good and bass is good too. But I can't say too much as I am playing it quite soft.
So my question is answered. High Zout preamps sound fine if we operate them within reason. All parameters must be known , load impedance , capacitance and drive voltage required. We can eliminate the cathode follower and reduce the number of active devices in the chain.
This preamp appears to get up to about 10 V rms before the distortion rises rapidly. I need less than 2 V rms to drive the power amp into clipping. Can't wait to check it out tomorrow.
Now I have to crash out. It's been a long day !
I tested the tube preamp in a hurry tonight. Zout is high and appears to be a bit higher than 8K ohms. Response with a 8.7K ohm load and 320pF capacitance is -0.6dB at 20Khz ( measured). Distortion with a CCS is 0.03% at 2 V rms ! With a resistive load at the anode the distortion is much higher. With a CCS I see practically only the second harmonic the higher orders being very low . With a resistive anode load the third fourth and fifth are also significantly present though they keep decreasing in amplitude. So CCS is the way to go.
I rigged it up with one power amp and I must say that it sounds very nice. I can't play it loud as it's midnight here. Tomorrow or Sunday will be the day for regular listening sessions. I don't hear any loss of HF and it doesn't sound 'bright'. Voice sounds very good and bass is good too. But I can't say too much as I am playing it quite soft.
So my question is answered. High Zout preamps sound fine if we operate them within reason. All parameters must be known , load impedance , capacitance and drive voltage required. We can eliminate the cathode follower and reduce the number of active devices in the chain.
This preamp appears to get up to about 10 V rms before the distortion rises rapidly. I need less than 2 V rms to drive the power amp into clipping. Can't wait to check it out tomorrow.
Now I have to crash out. It's been a long day !
Last edited:
Listening tests.
I spent 6 hours listening to the high Zout preamp.I used a Rotel RA-970BX integrated amp. This one is modified. The input goes directly from the input selector to the volume pot which goes directly to the power amp via a film cap. I removed the original Black Gate which was not as good as the film cap.
So it is just a power amp now. The bypassed preamp and tone control circuitry degrades the sound.
This is a very good sounding preamp. No apparent loss of highs and seems to be good right across the bandwidth. Great transients and clean HF.
Compared to an older preamp I had which has a CF output , this premp is much better. Voices sound really nice. The Rotel sounds better via the pre amp. This premp transforms the amp altogether. I've seen that sort of thing happen many times in other systems.
I kept all interconnects not longer than 2 feet each. There were several cables.
Cable from DVD player coax digital output to DAC coax SPDIF input. DAC analog output to passive Alps volume control. Cable from Alps to tube preamp and cable from preamp to power amp input. Three cables carrying the analog signal !
Now I need to compare this one with my 12B4A preamp which is in storage. I'll have to find a power supply for it.
The 12B4A also does not have a CF output stage. That preamp used to sound quite nice but the comparison is going to be interesting. The CF stage can be eliminated from the signal path if possible. Of course long interconnects can't be used .....unless one uses cables with very low capacitance .
Cheers.
I spent 6 hours listening to the high Zout preamp.I used a Rotel RA-970BX integrated amp. This one is modified. The input goes directly from the input selector to the volume pot which goes directly to the power amp via a film cap. I removed the original Black Gate which was not as good as the film cap.
So it is just a power amp now. The bypassed preamp and tone control circuitry degrades the sound.
This is a very good sounding preamp. No apparent loss of highs and seems to be good right across the bandwidth. Great transients and clean HF.
Compared to an older preamp I had which has a CF output , this premp is much better. Voices sound really nice. The Rotel sounds better via the pre amp. This premp transforms the amp altogether. I've seen that sort of thing happen many times in other systems.
I kept all interconnects not longer than 2 feet each. There were several cables.
Cable from DVD player coax digital output to DAC coax SPDIF input. DAC analog output to passive Alps volume control. Cable from Alps to tube preamp and cable from preamp to power amp input. Three cables carrying the analog signal !
Now I need to compare this one with my 12B4A preamp which is in storage. I'll have to find a power supply for it.
The 12B4A also does not have a CF output stage. That preamp used to sound quite nice but the comparison is going to be interesting. The CF stage can be eliminated from the signal path if possible. Of course long interconnects can't be used .....unless one uses cables with very low capacitance .
Cheers.
Yes a cathode follower is the typical solution to buffer a low impedance load with a high impedance source like a tube preamp.
However I'm trying to minimise the number of stages in the signal path without causing any nasty side effects.
About balanced connections . The problem posed by a low impedance load with capacitance with a high impedance source should be similar in a balanced or unbalanced system. The treble will still roll off . The transition frequency will depend on the RC value and as it is 1/(2*pi*R*C) , a lower C will raise the transition frequency and reduce the loss at audio frequencies.
Lowering the load impedance will cause gain loss and increased distortion. Increasing the capacitance ( of the load ) will cause increasing loss at audio frequencies and increase distortion at those frequencies.
However one can try and juggle circuitry and component values to get an acceptable result. I don't believe there is an optimum solution. Just one that is acceptable ...........if one wants it! Most people would probably scoff at it.
Cheers.
It's 11pm now and time to crash out. Goodnight.
However I'm trying to minimise the number of stages in the signal path without causing any nasty side effects.
About balanced connections . The problem posed by a low impedance load with capacitance with a high impedance source should be similar in a balanced or unbalanced system. The treble will still roll off . The transition frequency will depend on the RC value and as it is 1/(2*pi*R*C) , a lower C will raise the transition frequency and reduce the loss at audio frequencies.
Lowering the load impedance will cause gain loss and increased distortion. Increasing the capacitance ( of the load ) will cause increasing loss at audio frequencies and increase distortion at those frequencies.
However one can try and juggle circuitry and component values to get an acceptable result. I don't believe there is an optimum solution. Just one that is acceptable ...........if one wants it! Most people would probably scoff at it.
Cheers.
It's 11pm now and time to crash out. Goodnight.
I've just woken up and looked at the response. Just one !
RDF , it is a great situation to have spare OPT's in the junk box. However these are not collectible items here. Not now anyway and I doubt if we can get a good one. 15K:600 would be quite useful.
After doing some more testing I find the preamp behaving more like a current source than a voltage source! Is that why it sounds so different ? This probably needs some more exploration.
Anyone used current sources for audio ? Not CCS but current audio "source" ?
Cheers.
RDF , it is a great situation to have spare OPT's in the junk box. However these are not collectible items here. Not now anyway and I doubt if we can get a good one. 15K:600 would be quite useful.
After doing some more testing I find the preamp behaving more like a current source than a voltage source! Is that why it sounds so different ? This probably needs some more exploration.
Anyone used current sources for audio ? Not CCS but current audio "source" ?
Cheers.
For those who might be interested ...I ran some more tests.
1. I compared the direct DAC sound with the output of the preamp fed by the DAC output.
2. I listened on speakers and also a pair of HD580's.
3. Some of the music was Janis Ian -Breaking Silence , Pink Flyod -The Wall and - Best of Herb Alpert etc.
I adjusted levels so that they were matched to within 0.1 dB.
Switching between DAC and preamp output showed almost no difference on most music except in two areas . There is an audible reduction in extreme HF which is noticeable only in some parts of the music as they are not present everywhere. Sounds like brushed cymbals and some shimmering sounds are 'slightly' less bright. If you weren't looking for it you'd not notice it.
Playing pink noise you can clearly make out this 'slight' reduction of extreme HF.
Strangely the bass is mostly similar except that on some very tight deep bass sounds the preamp seems to have more 'energy' in the slam. It also appears to be slightly better defined. This however is noticeable only on some tracks.
The slight reduction in HF appears to affect the midrange ! Voices sound smoother particularly when they go very loud ! Is it possible that when the voices get that loud you get a lot of higher order harmonics which in this preamp gets rolled off ? Voice sounds fantastic when played at realistic levels. It's sometimes has a bit of an edge ( on some recordings) direct from the DAC.
The shocker is that the preamp Zout measured about 79 K ohms !
I goofed the sim. It actually worked out to be 71K ohms. Anyway it's been a very illuminating experiment.
I might add that on various experiments I did sometime ago I did find that extreme HF can get attenuated depending on the type of film cap you use. Here I used inexpensive Epcos polyprop caps. I'll be substituting them with Jensen PIO's and Mundorf caps to see what happens.
So high Zout isn't always a bad thing. One can get away with higher mu tubes than one wants !
Cheers.
1. I compared the direct DAC sound with the output of the preamp fed by the DAC output.
2. I listened on speakers and also a pair of HD580's.
3. Some of the music was Janis Ian -Breaking Silence , Pink Flyod -The Wall and - Best of Herb Alpert etc.
I adjusted levels so that they were matched to within 0.1 dB.
Switching between DAC and preamp output showed almost no difference on most music except in two areas . There is an audible reduction in extreme HF which is noticeable only in some parts of the music as they are not present everywhere. Sounds like brushed cymbals and some shimmering sounds are 'slightly' less bright. If you weren't looking for it you'd not notice it.
Playing pink noise you can clearly make out this 'slight' reduction of extreme HF.
Strangely the bass is mostly similar except that on some very tight deep bass sounds the preamp seems to have more 'energy' in the slam. It also appears to be slightly better defined. This however is noticeable only on some tracks.
The slight reduction in HF appears to affect the midrange ! Voices sound smoother particularly when they go very loud ! Is it possible that when the voices get that loud you get a lot of higher order harmonics which in this preamp gets rolled off ? Voice sounds fantastic when played at realistic levels. It's sometimes has a bit of an edge ( on some recordings) direct from the DAC.
The shocker is that the preamp Zout measured about 79 K ohms !
I goofed the sim. It actually worked out to be 71K ohms. Anyway it's been a very illuminating experiment.
I might add that on various experiments I did sometime ago I did find that extreme HF can get attenuated depending on the type of film cap you use. Here I used inexpensive Epcos polyprop caps. I'll be substituting them with Jensen PIO's and Mundorf caps to see what happens.
So high Zout isn't always a bad thing. One can get away with higher mu tubes than one wants !
Cheers.
Last edited:
Hi Andrew,
I'm not sure if you are talking about the sim or the actual measurement.
Sim: I apply a 1uA ac current source to the output terminal and measure the voltage at the output terminal. V/I gives me Zout. The input is shorted to ground via a 680 ohm resistor. For this sim I used Beige Bag V5 .
For the actual measurement I used a Clio system . All you need to do is connect the oscillator output to the output of the preamp and also connect this to the Clio input terminal. Then Run the 'impedance' application. The measurement is off scale on the monitor but the cursor does give a reading.The scale is capable of showing a max of 10K ohms on screen.
Alternatively I'd use a signal generator connected to the output of the preamp via a high value resistor and measure the voltage across the preamp output terminals. So I have two known voltages , one known resistor and one unknown impedance. Simple maths gives me the output impedance at the frequency that I have used. I can sweep the frequency and plot it out ( manually) . I could also possibly use ARTA which is not yet fully set up on my machine. The hardware is still being worked on. I haven't looked at LAud . It is setup on my Clio machine but isn't calibrated and I don't use it much. Maybe I should look at it again.
Cheers.
I need to migrate to ARTA because both Clio and LAud ( Fiji sound card ) need a PC with an EISA slot. I have another spare PC with EISA slots in case this one breaks down. I had four till not long ago and two broke down. Can't get spare parts at sensible prices. Better to move on to better and newer measurement software. ARTA is on my newer PC and I've used it a few times. I like it. They are working on additional features I think.
I'm not sure if you are talking about the sim or the actual measurement.
Sim: I apply a 1uA ac current source to the output terminal and measure the voltage at the output terminal. V/I gives me Zout. The input is shorted to ground via a 680 ohm resistor. For this sim I used Beige Bag V5 .
For the actual measurement I used a Clio system . All you need to do is connect the oscillator output to the output of the preamp and also connect this to the Clio input terminal. Then Run the 'impedance' application. The measurement is off scale on the monitor but the cursor does give a reading.The scale is capable of showing a max of 10K ohms on screen.
Alternatively I'd use a signal generator connected to the output of the preamp via a high value resistor and measure the voltage across the preamp output terminals. So I have two known voltages , one known resistor and one unknown impedance. Simple maths gives me the output impedance at the frequency that I have used. I can sweep the frequency and plot it out ( manually) . I could also possibly use ARTA which is not yet fully set up on my machine. The hardware is still being worked on. I haven't looked at LAud . It is setup on my Clio machine but isn't calibrated and I don't use it much. Maybe I should look at it again.
Cheers.
I need to migrate to ARTA because both Clio and LAud ( Fiji sound card ) need a PC with an EISA slot. I have another spare PC with EISA slots in case this one breaks down. I had four till not long ago and two broke down. Can't get spare parts at sensible prices. Better to move on to better and newer measurement software. ARTA is on my newer PC and I've used it a few times. I like it. They are working on additional features I think.
Last edited:
I think you want me to measure this. I can't do it with Clio as it's input impedance is 64K ohms. I currently don't have a buffer to load the preamp with 1 meg or even 79K ohms. The alternative is to use an mV meter . My wide bandwidth DMM isn't working ! Let me dig out my analog mV meter and see if I can do this sometime tonight.
Cheers.
Cheers.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.