Sorry, my bad. Q7 and Q8 are in fact the same current mirrors as in the Hiraga schematic. In fact, the two preamps are pretty much identical, signal path wise.
BTW, it seems like in this particular case, capacitors (electrolytics!) in the signal path are acceptable. I can't stop recalling how one of the contributors around expressed his disgust for my HPS 1.0 having a a film cap in the input stage signal path
P.S. The ML preamp is just another piece of ancient technology. Those transistors were considered "low noise" about 40 years ago. That's the reason why it can afford running at low Ic, Rbb is so large that it masks any current noise contribution. Such a preamp will fail the noise test with any modern 10ohm low output cartridge.
Last edited:
syn08,
grounded base is not a suitable topology for MM, where the input impedance needs to be set externally, by listening as the only way, the values can be up to 100kOhm.
An increased bias current improves almost all parameters of interest, in all topologies, like noise. In practice, compromise is always necessary, not forgetting the annoying thermal-dependent DC offset in grounded base, for instance.
grounded base is not a suitable topology for MM, where the input impedance needs to be set externally, by listening as the only way, the values can be up to 100kOhm.
An increased bias current improves almost all parameters of interest, in all topologies, like noise. In practice, compromise is always necessary, not forgetting the annoying thermal-dependent DC offset in grounded base, for instance.
That´s not a factual assertion. The equations are easily accessible.
Set aside MM is off topic here, that's nonsense anyway. Manufacturer are specifying the optimal load for MM. Of course, some GEB members can make some good coin by helping ignorant user in optimizing their sound using the "critical listening" method.
The common base and common emitter nondegenerated transconductances are identical. The common base input impedance is lower (my bad) and that's the real reason why you can't use high currents in the Hiraga stage. The input impedance is 1/gm. If this is to be set for 100 ohm, this would require a maximum collector current of 250uA. At this current level, you don't need special low noise devices, you can use any medium power double diffused bipolar from your junkbox, with the same crappy results.
Last edited:
I did. It is entirely correct. As usual, your blank statements have zero factual value.
In case you use a low impedance cardridge the input impedance can be much lower then 100 Ohms without harm. I use a Lyra Titan with an impedance of 5.5 Ohm and inductance of less then 0.5uH. It certainly works well into the MPP that has an input resistance of 25 Ohms in case i shunt the current mirrors with caps. I run the input transistors on 3.5mA each. Output transresistance at that point is 50mS for the MAT02.
A bit higher in the MAT03. Consider also that no current is "burned" in a shunt resistor but everything flows into the input transistors. With a high impedance design shunted with 100 Ohms you loose around 5.5% of current (with the Titan). Hardly much but maybe audible compared with say a 47 kOhm load just because the sound is not as load. Douglas Self claims that a volume change of 0.3dB is just audible.
I have not finished my survey into cardtridge loading but i will try to calculate the volume differences too. For a Denon DL 103 with 40 Ohms and more inductance a loading too low will roll of the treble because the inductance with interact with the low input impedance. I also learned from Malkolm Hawksford that after they decided on the input impedance of the Essex Equalizer they found out that samples sounded very different. Although they always used the same value resistor they found that different makes and brands sounded very different especially when they put the needle in the groove and listened to the mechanical and electrical noise. I think they settled on a tantalum resistor. Call me biased or stupid i trust that man and can hear it myself. Has anybody a scientific explanation to persuade Syn08 into listening to various makes of resistors ? i do not think so.
A bit higher in the MAT03. Consider also that no current is "burned" in a shunt resistor but everything flows into the input transistors. With a high impedance design shunted with 100 Ohms you loose around 5.5% of current (with the Titan). Hardly much but maybe audible compared with say a 47 kOhm load just because the sound is not as load. Douglas Self claims that a volume change of 0.3dB is just audible.
I have not finished my survey into cardtridge loading but i will try to calculate the volume differences too. For a Denon DL 103 with 40 Ohms and more inductance a loading too low will roll of the treble because the inductance with interact with the low input impedance. I also learned from Malkolm Hawksford that after they decided on the input impedance of the Essex Equalizer they found out that samples sounded very different. Although they always used the same value resistor they found that different makes and brands sounded very different especially when they put the needle in the groove and listened to the mechanical and electrical noise. I think they settled on a tantalum resistor. Call me biased or stupid i trust that man and can hear it myself. Has anybody a scientific explanation to persuade Syn08 into listening to various makes of resistors ? i do not think so.
i agree
if i would do a commercial phonostage it should have adjustable input impedance and capacitance. there are too many different designs on the market (cartridges)
you will be surprised that i often use a 10nF capacitor parallel to the common base stage.
it sounds excellent with the Titan in my setup. maybe the incredible speed of the common base stage is a problem to let in some hf that is not welcome.
if i would do a commercial phonostage it should have adjustable input impedance and capacitance. there are too many different designs on the market (cartridges)
you will be surprised that i often use a 10nF capacitor parallel to the common base stage.
it sounds excellent with the Titan in my setup. maybe the incredible speed of the common base stage is a problem to let in some hf that is not welcome.
2n4401 is not bad even today
to claim Douglas Self again in his book "Self on Audio" "exchanging the 2n4401 with something more modern like the 2sb797 brings only 1dB noise advantage in this circuit"
maybe not exactly his words but what i understand
JC i owned and still own original Mark Levinson equipment equipment and although we never met you are a great inspiration for me.
you can not imagine how strange and wonderful it was as a young man in the 70th to look and listen to such amasing equipment. it simply was in a class of it`s own besides original Threshold gear. thank you for triggering my interest in REAL quality.
to claim Douglas Self again in his book "Self on Audio" "exchanging the 2n4401 with something more modern like the 2sb797 brings only 1dB noise advantage in this circuit"
maybe not exactly his words but what i understand
JC i owned and still own original Mark Levinson equipment equipment and although we never met you are a great inspiration for me.
you can not imagine how strange and wonderful it was as a young man in the 70th to look and listen to such amasing equipment. it simply was in a class of it`s own besides original Threshold gear. thank you for triggering my interest in REAL quality.
I 'discovered' the 2N4401-4403 combination in 1968, while working in the Ampex Audio department, designing tape recorders. I had a Quan-Tech, and I used it. The first public mention of the 2N4403 was in a 1971 article by H.P. Walker about low noise design. You will find a footnote in there, with my name attached to it. This combination was the best until about 1978 or so. That is why Levinson used it. Syn08 needs to understand this. The effective Rbb' was about 40 ohms, very low for the time.
Last edited:
John:
Short of your usual personal comments, I agree with most of what you are saying. The only problem (not new, though) is that you are still stuck in the 70's. 40ohm Rbb is by today standards horribly high, and running bipolars at a quarter of mA to get 100ohm input impedance is not helping at all in noise performance. I haven't done any exact calculations, but the total noise of that common base stage is unnaceptable today. In fact, you could get the same or even better noise performance using any modern low noise JFET input opamp, not to mention an under 1nV/rtHz bipolar opamp.
Joachim:
I have never tried to go that low with the cartridge loading, but certainly a Lyra Titan is not for everybody's pockets. To build a preamp that would accomodate a wide number of MCs, I think 25ohm is way to low.
Considering 100ohm as input impedance, then Doug's comments make a lot of sense. At 250uA collector current, the current noise dominates the total noise, hiding most of the Rbb effects. Therefore using low noise transistors doesn't help much. Remeber that the current noise decreases with the square of collector current, therefore between 250uA and e.g. 1mA there's a factor of 16 reduction in current noise! Obviously using 25ohm input impedance, as you did, helped reducing the noise.
Yes, it would be pretty hard to persuade me there's an audible difference between tantalum and regular metal film resistors.
Short of your usual personal comments, I agree with most of what you are saying. The only problem (not new, though) is that you are still stuck in the 70's. 40ohm Rbb is by today standards horribly high, and running bipolars at a quarter of mA to get 100ohm input impedance is not helping at all in noise performance. I haven't done any exact calculations, but the total noise of that common base stage is unnaceptable today. In fact, you could get the same or even better noise performance using any modern low noise JFET input opamp, not to mention an under 1nV/rtHz bipolar opamp.
Joachim:
I have never tried to go that low with the cartridge loading, but certainly a Lyra Titan is not for everybody's pockets. To build a preamp that would accomodate a wide number of MCs, I think 25ohm is way to low.
Considering 100ohm as input impedance, then Doug's comments make a lot of sense. At 250uA collector current, the current noise dominates the total noise, hiding most of the Rbb effects. Therefore using low noise transistors doesn't help much. Remeber that the current noise decreases with the square of collector current, therefore between 250uA and e.g. 1mA there's a factor of 16 reduction in current noise! Obviously using 25ohm input impedance, as you did, helped reducing the noise.
Yes, it would be pretty hard to persuade me there's an audible difference between tantalum and regular metal film resistors.
Yes, it would be pretty hard to persuade me there's an audible difference between tantalum and regular metal film resistors.
Your loss I guess. You obviously have one system that you listen too often. Try and see if you can hear the differences. I am not a SS guy, but I am sure someone can tell you a good location to try to hear these differences. A MC input stage should work. If you can not hear differences in capacitors, resistors might be tough. The truth is there in front of you. Just allow it to happen.
One of my early mentors was a Fellow at large electronics company. He showed me the sonic differences in many items that have to do with passive components in electronic circuits.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.