Talking about noise and avalanche breakdown, there is a chapter “AVALANCHE BREAKDOWN AND NOISE” in Motchenbacher/Connelly that gives the following warning that may be of interest:
So far so good, but I just ordered a automatic component tester – Peak Atlas DCA 55 – that lets you “connect your component any way round and press the test button”, i.e. it searches for the correct terminals of the device. Now, the peak test voltage is supposedly 5.1 Volts, which is exactly where the EB breakdown of a transistor occurs according to the datasheets. I am just beginning to wonder how safe it is to use this type of tester. But again, I may be worrying to much (though I would not want to "tune" the noise performance of my transistors)
So far so good, but I just ordered a automatic component tester – Peak Atlas DCA 55 – that lets you “connect your component any way round and press the test button”, i.e. it searches for the correct terminals of the device. Now, the peak test voltage is supposedly 5.1 Volts, which is exactly where the EB breakdown of a transistor occurs according to the datasheets. I am just beginning to wonder how safe it is to use this type of tester. But again, I may be worrying to much (though I would not want to "tune" the noise performance of my transistors)
This is what usually leaves me in a dazed state and also where the communication breaks down: In a MC pre, you say 1nV/rtHz noise is not an issue 'cause most people will live with it; you say open loop is fine, 'cause distortions are anyway not audible; and then you praise non-magnetic resistors, and other such, for their superior sound.
I (and I dare to say, anyone of my EE breed, not involved in the audio industry) would tell you that there are 1001 reasons, some of them many orders of magnitude larger, impacting the sound in your designs, before you can talk about (non)-magnetic resistors.
Under these circumstances, I doubt I can myself suggest anything for your designs, that you would consider as a "sonic improvement". All I can do is to wonder and to query your design decisions, in an attempt to find correlations (if any) with your "sonic results".
Talking about noise and avalanche breakdown, there is a chapter “AVALANCHE BREAKDOWN AND NOISE” in Motchenbacher/Connelly that gives the following warning that may be of interest:
So far so good, but I just ordered a automatic component tester – Peak Atlas DCA 55 – that lets you “connect your component any way round and press the test button”, i.e. it searches for the correct terminals of the device. Now, the peak test voltage is supposedly 5.1 Volts, which is exactly where the EB breakdown of a transistor occurs according to the datasheets. I am just beginning to wonder how safe it is to use this type of tester. But again, I may be worrying to much (though I would not want to "tune" the noise performance of my transistors)
Different avalanche than collector "weak" avalanche which is probably what is refered too previously. Increased noise due to this is complex and device dependent. A few days with the literature and some bench work could yield a useful list of devices that look good at high Vcb, but are no longer low noise. Some might keep this a trade secret even.
Avalanching a Veb generates hot carriers that can and do permanently or semi-permanently damage the junction making it (more not less) noisey.
Equipment in general would not survive the temperatures needed to anneal this damage.
Last edited:
Hello Syn08 !
When i use your formular with the 12 BC550 and my 5.5Ohm cartridge i end up with 117ma, 9.7mA for each transistor. That is a bit unpractical. On the other hand the Rbb of the BC550C does not change much with collector current. It has 170 Ohm at 0.1mA, 163 Ohm at 0.5mA, 160 Ohm at 1mA and 130 Ohm at 10mA. So something between 0.5 and 3mA could be a compromise. Anyway, i will not eleborate more on that circuit. I went to far. A row of 4 2N4401 or 2N4402 is a better idea i think for a cheap Headamp.
I will recalculate the values here and present a circuit that works. For the cautios: it presents a DC current to the cartridge at turn on and it works best with low impedance MCs. I praxis i have never burned a cartridge though.
When i use your formular with the 12 BC550 and my 5.5Ohm cartridge i end up with 117ma, 9.7mA for each transistor. That is a bit unpractical. On the other hand the Rbb of the BC550C does not change much with collector current. It has 170 Ohm at 0.1mA, 163 Ohm at 0.5mA, 160 Ohm at 1mA and 130 Ohm at 10mA. So something between 0.5 and 3mA could be a compromise. Anyway, i will not eleborate more on that circuit. I went to far. A row of 4 2N4401 or 2N4402 is a better idea i think for a cheap Headamp.
I will recalculate the values here and present a circuit that works. For the cautios: it presents a DC current to the cartridge at turn on and it works best with low impedance MCs. I praxis i have never burned a cartridge though.
It's not "my" formula, it's a classic but of course, that's to much. A good sign that you are using a very low impedance cartridge, therefore you may neglect the current noise - but you can't expect everybody to use a $5k Lyra Titan MC cartridge!
Actually, you got it pretty well - 2N5401 and 2N5551 are about the lowest noise TO92 devices currently in production in the USA, they are much better than BC550 and BC560. They have Rbb' around 100ohm. There's a table with measured Rbb's on my web site: Low Noise Design Schematics
Rbb' is barely supposed to vary with Ic.
Same thing, really. Impact ionisation mechanism in both cases, increases as McIntyre with k=0.2. The absolute cross sections, multiplication and threshold are very device and process dependent, but for common epitaxial devices it can be assumed the excess noise matters at Vce>0.5*Vcb0. Another good reason to close the loop, to keep Vce close to bias, can usually be about 5V.
Last edited:
Echo simplife
Here is a circuit that should work. I choose the 2N4403. Adjust the fet with the trimmer until 4mA are flowing in the 750 Ohm resistor. A fet with Idss of around 8mA is fine. I found that the current that flows in the current mirror triples in the input transistors. This is because of the 100 Ohm resistor. T6 can be omitted. After adjusting the bias DC has to be cancelled with the trimmer to the positive supply. Higher Bias could be tried out, that needs some other resistor values. I use 1mA here because that is the value that Douglas Self uses for the 2N4403 and his circuit has a reputation for low noise.
Here is a circuit that should work. I choose the 2N4403. Adjust the fet with the trimmer until 4mA are flowing in the 750 Ohm resistor. A fet with Idss of around 8mA is fine. I found that the current that flows in the current mirror triples in the input transistors. This is because of the 100 Ohm resistor. T6 can be omitted. After adjusting the bias DC has to be cancelled with the trimmer to the positive supply. Higher Bias could be tried out, that needs some other resistor values. I use 1mA here because that is the value that Douglas Self uses for the 2N4403 and his circuit has a reputation for low noise.
Now you need to add local power supply regulators/buffers and a cascode. Otherwise, this thing has virtually zero PSRR. If you plan to feed via batteries, it's probably good as is. I would though suggest at least 5mA. and decrease the load resistor for the same gain.
Step back for a moment and look at your design. Do you like the caps in the signal path? Wait until M. Fremer will find out about
Syn,
I feel you are critiquing something you don't fully understand. You have seen a schema of the RTP phono stage, but I suspect know nothing of the overall design and full specs.
I suggest you check out:
http://www.vacuumstate.com/fileupload/RTP3D_brochure_b.pdf
and see that there has been some serious engineering taken place. And look carefully at the distortion numbers and graphs on Page 4. If you can prove you can hear the difference in distortion between the RTP3D and one of your designs, I'll buy you lunch anywhere you choose.
You critiqued my use of 22ohm stopper R's in the MAT02 inputs, saying they add noise, and I replied it stops oscillation. You said I should be able to fix that with layout - maybe you could, but after two weeks of work it was easier to add the R's. The preamp has significant gain at 3MHz, and the phono and line stages are in one box and all inputs and outputs are at the same end.
Last point, I assure you I can clearly hear the difference between regular magnetic end cap R's and the Dale/Vishay non magnetic R's - we heard it first in our SACD player Upgrade module, and proved it in our tube gear. For someone who mostly uses the PC to simulate circuits, how can you critique my perceptions? Would you not use the best possoble parts, costing allowing?
Or am I just wasting my time talking with you?
Regards, Allen
I feel you are critiquing something you don't fully understand. You have seen a schema of the RTP phono stage, but I suspect know nothing of the overall design and full specs.
I suggest you check out:
http://www.vacuumstate.com/fileupload/RTP3D_brochure_b.pdf
and see that there has been some serious engineering taken place. And look carefully at the distortion numbers and graphs on Page 4. If you can prove you can hear the difference in distortion between the RTP3D and one of your designs, I'll buy you lunch anywhere you choose.
You critiqued my use of 22ohm stopper R's in the MAT02 inputs, saying they add noise, and I replied it stops oscillation. You said I should be able to fix that with layout - maybe you could, but after two weeks of work it was easier to add the R's. The preamp has significant gain at 3MHz, and the phono and line stages are in one box and all inputs and outputs are at the same end.
Last point, I assure you I can clearly hear the difference between regular magnetic end cap R's and the Dale/Vishay non magnetic R's - we heard it first in our SACD player Upgrade module, and proved it in our tube gear. For someone who mostly uses the PC to simulate circuits, how can you critique my perceptions? Would you not use the best possoble parts, costing allowing?
Or am I just wasting my time talking with you?
Regards, Allen
Allen,
You have conveniently answered only the questions you felt comfortable with, avoiding some and without specifying the conditions. Not that it really matters, but care to specify distortions to allow an apple to apple comparison I specified the distortions at 20KHz and 20Vp output, you did at 1Vrms out and (not really sure, the graph is unclear) 1KHz. Though, no doubt a nice product and I'm sure it also sounds great.
Otherwise, if you put it this way, then yes you are wasting your time with me. Good luck listening to the non-magnetic resistors and make sure you better advertise this technological breakthrough finding, together with the "atomic power station quality primary signal path capacitors" which I'm sure also make a huge difference.
Maybe, but statements like in Allen's brochure (quoted above) are extraordinary, therefore the burden of proof, as having any support other than marketing tactics, is on the authors.
I've said this over and over again: I'm fine with statements like "according to our marketing studies, this spec [THD, noise, you name it] is good enough for 99.9% of our customers". I'm not fine with inflating the price by using expensive components, only for the purpose of filling in the product brochure with technobabble **. In this sense, I fully agree that open loop MC headamps are feasible and, as long as the vinyl distortions are not better than 0.1% anyway, they may offer all that customers are looking for, but I'm not fine with claims there's anything sonically fundamentally superior in such a configuration. Or that BJTs cascoded with a tube have some undisclosed superiority. Or that non-magnetic resistors make a difference here. Or about "atomic power station quality primary signal path capacitors".
Prove it, or keep it at the marketing level only. For the rest, see the Cable thread.
I use the LC audio low noise supply with 25mH coils in the supply pass to the ECHO with 4 BC550c. Hum is not audible, ear on speaker. As far as i can see you syn08 use also one coupling capacitor at the output. I do not like a lot of coupling capacitors in the chain because they take away some micro detail in the extrem treble. There are some coupling caps that are very transpatent though. I use Riva-Evox PHE450 and Vishey-Röderstein MKPs that are not super expensive. Michael will have no problem with coupling caps if he likes the sound as far as i know him and has given praise to cap coupled designs. Sometimes a cap adds safety or is an easy solution in a circuit with very high DC gain like my inductive shelve RIAA. The Echo is not state of the art in any way but sounds very good to my own surprise. It also costs very little so i thought it is a good beginners circuit for people that whould like to explore transimpedance solutions. I will try with higher bias but will not eleborate the circuit any further. I will now concentrate to find a way to DC couple the MPP Low Z to the inductive shelve RIAA and then develop a High Z version with fets. I have also an idea to lower noise in the Low Z MPP and will ultimately go back to fully balanced, also in the RIAA. That soution has a 3dB noise penalty of cause. My goal is 0.5nVQHz. I think with "normal" sensitivity speakers in the 86 to 91dB range that is fine to be not audible in the listening seat. Any improvement is wellcome. I said it before, i see no reason why your more aleborated phonostages should not sound exemplary. I just have not heard them in my home and you already found out that for me subjective apraisal is more importand then exeptional measurements. P.S. : my commercial phonostage for Spiral Groove is totaly DC coupled from input to output.
I connected the first time the MPP Low Z to the Inductive Shelve RIAA. For the time being
AC coupled with Rifa PHE450 2.2uF and 0.1 Uf Röderstein MKP bypass. How does is sound compared to the Echo? Well the Echo is very transparent and has a lot of speed but the MPP Low Z is more natural and easy on the ears. I am very happy with that combination and will listen more tomorrow.
AC coupled with Rifa PHE450 2.2uF and 0.1 Uf Röderstein MKP bypass. How does is sound compared to the Echo? Well the Echo is very transparent and has a lot of speed but the MPP Low Z is more natural and easy on the ears. I am very happy with that combination and will listen more tomorrow.
Sorry to butt into the thread (I've been reading, but not posting) however...
I am confused as to why speaker sensitivity would matter to hearing noise in a phono stage. Noise in the power amp, sure, because there wouldn't be any attenuation of that noise.
But the phono stage is ultimately going to be amplified then attenuated (volume control) to some nominal listening level. Whether that level is produced by speakers that are 86dB@2.83V or 98dB@2.83 volts the acoustic level is going to be the same - E.G. 84dB average. Noise is going to be XdB below that, no matter what.
Or am I missing something?
I am confused as to why speaker sensitivity would matter to hearing noise in a phono stage. Noise in the power amp, sure, because there wouldn't be any attenuation of that noise.
But the phono stage is ultimately going to be amplified then attenuated (volume control) to some nominal listening level. Whether that level is produced by speakers that are 86dB@2.83V or 98dB@2.83 volts the acoustic level is going to be the same - E.G. 84dB average. Noise is going to be XdB below that, no matter what.
Or am I missing something?
Last edited:
It doesn't matter to most grown ups. Younger "audiophiles" though sometimes turn the volume up to the max, glue their ears to the tweeters and demand that no hiss is heard. Playing some music at that very moment would probably teach them well.
Problem with me doing that is that the overwhelming noise source is the thermal noise of the cartridge, and that's not something a phono preamp can do anything about.
Because of the power summing of uncorrelated noise contributors, as long as the preamp has a S/N 6dB below the cartridge, its contribution is negligible. That also goes for the RIAA compensation- I hear what Joachim is saying about Johnson noise from the inductor, but really, an RC network will not degrade the overall S/N if it's designed competently. For my own preamp, it causes less than a 0.1dB change in the noise, and doesn't suffer from the severe non-idealities of real-world inductors (and is a boatload cheaper, too).
Because of the power summing of uncorrelated noise contributors, as long as the preamp has a S/N 6dB below the cartridge, its contribution is negligible. That also goes for the RIAA compensation- I hear what Joachim is saying about Johnson noise from the inductor, but really, an RC network will not degrade the overall S/N if it's designed competently. For my own preamp, it causes less than a 0.1dB change in the noise, and doesn't suffer from the severe non-idealities of real-world inductors (and is a boatload cheaper, too).