How would you say parasitics compare with the "birds nest" vs PCB construction? Performance might get worst going to a PCB?
Yes averaging can lower dynamic resolution but only to a certain point.
0.3nVqHz seems to be the boundary at the moment at least at room temperature with silizium transistors
one posibility whould be gallium-arsenide cooled down low maybe with a peltier element
but the -140dB/U i get consistently with a variety of designs is quiet enough for me.
if an ultra low noise design whould sound any better is open to debate anyway.
i am not a great layouter so this work is done by another person for me.
he is so good that i get tears in my eyes.
of cause the physical properties of the pcb material have an influence.
for best sound in low level circuits i prefer air dialectric construction but there is very good pcb material availlable say from teflon or other composures albeit at a premium.
i do not have enough experience with different materials but listeng tests (and measurements) of the same circuit on different materials are on the agenda.
Here the Goldstandart at -30dB/U
Measurements 4th Installment
Here come the Measurements at -20dB/V
from left to right :
Goldstandart, MPP, Paris
Very interesting design - thank you very much indeed for puplishing it. Allow me a couple of stupid questions: What is the impedance of the input stage as shown and how much AC gain does it have (or what range can it have)?
By the way, I vagely remember a review of a phonostage from Audio Physic in a German hifi magazine many moons ago - it was highly appraised. Is this by any chance similar to the current design? I seem to remember some mentioning of a transimpedance stage or somthing of the kind.
The gain of the input stage alone is x20 plus x3 from the first stage of the Intrumentation amp without the 318usec and 3180usec RIAA in place with a cardridge of 5 Ohm impedance. the measurements i have shown here reflect that situation. So gain is 36dB.
it is 10dB higher then the Paris that has 26dB. So in praxis you do not have to crank the volume as much as with the Paris winning some gain in noise performance. The second stage of the Instrumentation amp will ad 24dB of gain at 1kHZ and 44dB of gain in the bass region to total gain is 60dB with compleet RIAA in place. This can easyly be changed (raised) by lowering R8 ore/and lowering R11,12 in equal measure.
In fact if i whould do the circuit all over again i whould distribute the gain between the input stage of the instumenation amp and the differential output stage more evenly to unburden the outputstage. With 24dB of gain at frequencies of 1kHz and higher the frequency response of the last stage is still several MHz because i use an OPamp with exeptionally high Gain-Bandwidth product but distributing the gain more evenly whould result in similar speed and distortion in both stages. The reason i came to this dimentioning was that i did not have the right value resistors on hand when i build the prototype. Still the contribution of the Intrumentation amp to distortion is very little. Most distortion comes from the input stage that has no voltage feedback so i wil give that stage more attention too.
Concerning the Audio Physic Phonostage it was a transimpedance design too but single ended not balanced and not parallel symmetric. The sound of that stage is still good today excelling in distiguished handling of dynamics. The problem of that stage was that it had automatic trigering and needed an impuls to switch on for example when you lower the cardridje to the record. With very low output cardridges like for example the Jan Allerts it could happen that it switched of in soft pasages giving rise to a lot of frustration by some users. the reason for that decision was the way we handled the battery supply. We did not want that the circuitry could be left on in case you forgot to switch it off to give the batteries more usefull life. some kind of weak KI, it was quite tempramental.
Input impedance : i run the input transistors of the MPP prototype on 2.6mA giving a dynamic input impedance of some 230 Ohms on both legs. So the plus input has 230 Ohm to ground and the minus input has 230 Ohm to ground dynamically. this does not reflect the noiseimpedance that is somewhat (much) lower at ca 30 Ohm. Shunting the bias transistors with 2200uF electrolytics halfs that values and gives a noiseadvantage of 3dB. I am just building the other channel with that option to compare. I use very good polymer electrolytics with only 8 milli ohms impedance at 100kHz so i hope they do no harm.
To syn08 and others that may be concerned about dumping current into the cardridge.
The MPP does not do that much because it is symmetrical and uses PNP and NPN types in equal measure on both inputs. yes i know the MAT03 has much less current gain but that is not a real problem in this design.
by the way i will try to advance the bias current to maybe 3.5mA. The MAT02 hits the sweetspot at 3mA but the MAT03 likes some more. I will optimise and discuss the bias currents during this thread in more detail.
More information i have gathered about dumping current into the cardridge : the generator will be pushed out of the middle position a bit not much different from raising the downpreasure. So second harmonic rises because that is an asymmetric distortion making the sound a bit more "full and heavy". Lowering the downpresure a bit can compensate for that giving a slightly lighter balance. I adjust downpreasure by ear anyway at least the microadjustment after i have reached satifying measurements.
RIAA and alternative second gainstage
I post a measurement of the RIAA curve i got from the prototype.
It shows a resolution of 1dB. The 1dB error at 20 Hz is due to the fact that i did not have a 318kOhm resistor around and used 3 parrallel 1MegOhm instead so this is easy to avoid. The result shows that the basic mathematics are correct.
As promissed i show also 2 alternative discreet circuits that can be used in case you do not like opamps in the signal chain. They contain the 318usec and 318ousec timeconstans implemented again as a transimpedance circuit.
They do the same job as the Intrumentation amp (actually not because they are conceptually much simpler) and are balanced in and out.
They can be further improved by bootstrapping so can the inputstage. i am not a great bootstrapper so ideas are very wellcome. i understand the basic idea but actually i never used that concept.
Thoose cicuits have to be finished of with two seperate unity gain buffers so one of my nexts post will show some options in various states of perfection and sofistication.
Next time i will send you measurements of the finished MPP prototype included the second and third time constants in full monty.
It occurred to me that two buffers are not needed if the output has not to be balanced two buffers would have an offset that needs additional attention if coupling capacitors are not on the wish list.
See my diagram where the Fet gain stage is finished of with a diamond buffer Hawksford style and a servo. I do not yet know where to connect the servo ideally. Give me some time.
When you look at the buffer it has a twist that the collectors of the input transistors are not connected to the power supply but to the emitters of the output transistors giving better performance with minimal complication. This is very typical for Malkolms thinking. Consider also that this circuit is over 20 years old. I had not even heard about a diamond buffer at that time.
Here are some simple use full buffers. The next post will show some more advanced circuits.
Performance of those circuits can be surprisingly good in terms of bandwidth
and distortion though they need a good power supply.
For the OP amp buffer i would use a current feedback design. They have better drive and speed then voltage feedback designs.
Be shure to use the correct value feedback resistor because stability in a current feedback OP is dependent on that resistor.
A short circuit from output to pin 2 does not work here.
|All times are GMT. The time now is 11:02 AM.|
Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio