fter 4 hour debugging yesterday(there was a contact-problem between Q3 and Q6 due to resoldering the Q3) my
Aleph Ono Prototyp now works well. The sound on my Rega Planar 3/2000 with
Elys (MM) is really detailed and IMHO the
rhythmic drives sounds very well (i have
heard Supertramp "brother where you bound and Bryan Ferrys "Boys and girls".
Ok some technical datas:
First i am using a torridal Transformer from
Thel-Audio Germany with nominal 120 VA.
I get on the PCB without any loads +/-
33,7 V after regulation and filtering and
42,6 V at the Relais-Outputs.
Due to the matching of the 4 2SK170 at the MC-Preamp-Stage i get over the Source-Resistors (R45 to R48) the following voltages:
R45Left = 90,3 mV; R45Right=94,8 mV
R46Left= 91,4 mV ; R46Right= 94,3 mV
R47Left= 89,6 mV; R47Right=93,7 mV
R48Left=90,1 mV; R48Right=97,6mV
Really nice values due to the carfully
matchings
)
Other interressting Voltages
On R67 i got on the Left side 1,061 V and on the right 1,066V
On R53Left=1,064V and right 1,062V
On R65Left= 613mV and right 632mV
On R64Left=915 mV and right the same
value.
On R57Left 5,94V and Right 5,90V
The Justage on the Poti R25P is a little bit difficult. I have noticed at the first
10 Min or so a slight drift to higher voltages (+ 10mV) so i decided to trim
it at 350 mV after 30 Min. I have also noticed that the MOSFET Q7,Q9,Q18,Q19 and especially the 9110D become hot (subjective i would say 50 °C ??). I think to put some adittionally small IC-heatsinks on the
MosFETs. I know from PQuadrat that normaly the MOSFET dont need heatsinks on the Aleph Ono but this would for me more safe and i hope that
the above voltage-drift on R28 and R27
would be more stable ?
Important for me is that on one Channel
the voltage over R27 is nearly (+/- 1-2 mV) the same as over R28 measured at the same time.
For cases i have found a nice reseller
on www.19zoll.com for german user.
I think to order a case with 2HE from this reseller. The sells now also on end-
users through there online shop. But the
2HE-Cases are available at KW 25/2003.
When its finished in this case i will post
some photos.
Aleph Ono Prototyp now works well. The sound on my Rega Planar 3/2000 with
Elys (MM) is really detailed and IMHO the
rhythmic drives sounds very well (i have
heard Supertramp "brother where you bound and Bryan Ferrys "Boys and girls".
Ok some technical datas:
First i am using a torridal Transformer from
Thel-Audio Germany with nominal 120 VA.
I get on the PCB without any loads +/-
33,7 V after regulation and filtering and
42,6 V at the Relais-Outputs.
Due to the matching of the 4 2SK170 at the MC-Preamp-Stage i get over the Source-Resistors (R45 to R48) the following voltages:
R45Left = 90,3 mV; R45Right=94,8 mV
R46Left= 91,4 mV ; R46Right= 94,3 mV
R47Left= 89,6 mV; R47Right=93,7 mV
R48Left=90,1 mV; R48Right=97,6mV
Really nice values due to the carfully
matchings
)Other interressting Voltages
On R67 i got on the Left side 1,061 V and on the right 1,066V
On R53Left=1,064V and right 1,062V
On R65Left= 613mV and right 632mV
On R64Left=915 mV and right the same
value.
On R57Left 5,94V and Right 5,90V
The Justage on the Poti R25P is a little bit difficult. I have noticed at the first
10 Min or so a slight drift to higher voltages (+ 10mV) so i decided to trim
it at 350 mV after 30 Min. I have also noticed that the MOSFET Q7,Q9,Q18,Q19 and especially the 9110D become hot (subjective i would say 50 °C ??). I think to put some adittionally small IC-heatsinks on the
MosFETs. I know from PQuadrat that normaly the MOSFET dont need heatsinks on the Aleph Ono but this would for me more safe and i hope that
the above voltage-drift on R28 and R27
would be more stable ?
Important for me is that on one Channel
the voltage over R27 is nearly (+/- 1-2 mV) the same as over R28 measured at the same time.
For cases i have found a nice reseller
on www.19zoll.com for german user.
I think to order a case with 2HE from this reseller. The sells now also on end-
users through there online shop. But the
2HE-Cases are available at KW 25/2003.
When its finished in this case i will post
some photos.
Hi Myxin,
i'am waiting for Peter's PCB's to build my own Ono. In advance,
I have done many psice simulations about the circuit and found something wrong in your measurements.
Matching the 2SK170 for same IDSS one thing, but getting the right working point another. Anyway, your matching is really good!
Question : Did you match teh FETs with or without source resistor ?
My oppinion is, that any 2SK170 in the mc-preamp has to work at about 2mA idle current (sum = 8mA).
If you use 22 Ohm source resistors, your measurement tells me, that one FET works at about 4 mA idle current. ( sum = 16mA)
I thing it is too much for the cascode (Q14) while the gain select jumper for R66 isn't installed. At sum of 16 mA idle current, normaly about 16 Volt should drop across R66+R40.
Result : Uc of Q14 = 14Volt. Otherwise the cascode is biased with Ub = 15 Volt. Theoreticly Ucb of Q14 is negativ !?!?
I don't thing that Nelson Pass is going to design amplifiers which can achieve conditions like that one !
toshiba
i'am waiting for Peter's PCB's to build my own Ono. In advance,
I have done many psice simulations about the circuit and found something wrong in your measurements.
Matching the 2SK170 for same IDSS one thing, but getting the right working point another. Anyway, your matching is really good!
Question : Did you match teh FETs with or without source resistor ?
My oppinion is, that any 2SK170 in the mc-preamp has to work at about 2mA idle current (sum = 8mA).
If you use 22 Ohm source resistors, your measurement tells me, that one FET works at about 4 mA idle current. ( sum = 16mA)
I thing it is too much for the cascode (Q14) while the gain select jumper for R66 isn't installed. At sum of 16 mA idle current, normaly about 16 Volt should drop across R66+R40.
Result : Uc of Q14 = 14Volt. Otherwise the cascode is biased with Ub = 15 Volt. Theoreticly Ucb of Q14 is negativ !?!?
I don't thing that Nelson Pass is going to design amplifiers which can achieve conditions like that one !
toshiba
I have finished my Ono too. The case I used is splitted into a small compact one (from Thel ALG2055) for the preamp and a simple one for the PSU which can placed anywhere in the rack. The connection between them is a simple but relatively thick 25-wire computercable, the plugs 25pin D-type. The PSU itself has a small 1.5VA stand-by transformer which provides a voltage for a relay that switches the main transformer.
The values of the current through the MC input FET's is nearly the same. Also the problem that the supply voltage of these FET's ist very low (about 6V) if the Jumper which shorts one of the 499R resistors is not fittet. With the jumper fittet the voltage is about 12V.
I did not fit the components for the inverter because I only need a asymmetric output. So the regulator transistors in the PSU keep only handwarm.
Regards
Dieter
The values of the current through the MC input FET's is nearly the same. Also the problem that the supply voltage of these FET's ist very low (about 6V) if the Jumper which shorts one of the 499R resistors is not fittet. With the jumper fittet the voltage is about 12V.
I did not fit the components for the inverter because I only need a asymmetric output. So the regulator transistors in the PSU keep only handwarm.
Regards
Dieter
Hi Dieter,
nice that you has finished your Ono. Hope I receive my PCBs from Peter this week.
If it is possible, do a hearing session with MC-Input and R66 Gain-Select-Jumper not installed. I'am sure, that the amp sounds more than worse !
The same answer to you.
It dosen't matter wheter the jumper is installed or not. The Emitter-Voltage of Q14 (Drain-Voltage of the 2SK170s) has to be "constant" !!!
The idle current of any of the 2SK170 must not exceed more than 2.5mA (sum of 4 = max. 10mA). Otherwise the cascode didn't work correctly if the jumper isn't installed !!!
toshiba
nice that you has finished your Ono. Hope I receive my PCBs from Peter this week.
If it is possible, do a hearing session with MC-Input and R66 Gain-Select-Jumper not installed. I'am sure, that the amp sounds more than worse !
The same answer to you.
It dosen't matter wheter the jumper is installed or not. The Emitter-Voltage of Q14 (Drain-Voltage of the 2SK170s) has to be "constant" !!!
The idle current of any of the 2SK170 must not exceed more than 2.5mA (sum of 4 = max. 10mA). Otherwise the cascode didn't work correctly if the jumper isn't installed !!!
toshiba
>The idle current of any of the 2SK170 must not exceed more than 2.5mA (sum of 4 = max. 10mA). Otherwise the cascode didn't work correctly if the jumper isn't installed !!!
The current through the FET's (BL-Type) is about 4mA, which is realy too high. Maybe for this circuit with the 22R Resistors BL-type ist the wrong type, GR types has a lower current and should be OK here.
I think I will order some of the GR and check out what happens.
Does anybody know which type is used in the original Aleph ONO?
The other way is to increase the restistor from 22R to 100R to get a current of 2mA (checked it out with a spare FET in the test circuit).
Regards
Dieter
The current through the FET's (BL-Type) is about 4mA, which is realy too high. Maybe for this circuit with the 22R Resistors BL-type ist the wrong type, GR types has a lower current and should be OK here.
I think I will order some of the GR and check out what happens.
Does anybody know which type is used in the original Aleph ONO?
The other way is to increase the restistor from 22R to 100R to get a current of 2mA (checked it out with a spare FET in the test circuit).
Regards
Dieter
Hi toshiba,
you are right,that for the input stage the iddle current must be lower than 16 mA , when the jumper across r 66 isn ' t installed.then the cascode doesn ' t work properly.
i am still wondering that in the circuit the Voltage at the Drain is about 15 V with or without the jumper.Perhaps you have to increase the rail to higher numbers , or to decrease the iddle current to proper value.
what Pspice Version do you have with simulation for the 2 SK 170 ?
you are right,that for the input stage the iddle current must be lower than 16 mA , when the jumper across r 66 isn ' t installed.then the cascode doesn ' t work properly.
i am still wondering that in the circuit the Voltage at the Drain is about 15 V with or without the jumper.Perhaps you have to increase the rail to higher numbers , or to decrease the iddle current to proper value.
what Pspice Version do you have with simulation for the 2 SK 170 ?
Hi hsort,
Hi LSV-Dieter, nice to meet you here!
Don't look at possible literal mistakes in the schematics of the Ono! Look at the circuit and try to understand it.
If the output-voltage of the MC-sub-regulator is really 26 Volt,
the cascode Q14 will be biased at it's base with about 13V.
So the Drain voltage can't be 15V !
Another (possible) mistake is "SHORT for -4db". Halving the gain-resistors (R66+R40) of the first stage normally gives -6dB. (Ok, I don't look at the following load-resistance)
And if you take a look here and see what Nelson Pass prints on his boards : "Remove for high gain + 3dB"
What is true ???
hsort, why to increase the rail, only why there is mark "15 VOLTS". I think 26 Volt is more than enough to get the amp work.
Yes, to decrease the idle current by increasing Rs is one way.
But what's about the gain (30dB) and those one, whose output of the MC-System is equal or less than 100uV. Good input value to the MM is about 10 mV.
You can calculate the gain in a simply way :
4 * (R66 + R40) / (RS + 1/gm)
"RS" is value of the source resistor (22) and "gm" the transconductance of the FETs (with assumption that all FETs have the same gm). Look at the datasheet of the 2SK170. Typical value of gm is 22mS. With R66 shorted, you get about 30 dB.
Sorry I lost the way !
Why are Pass amps so expensive and the circuits so simple ?
I think one of the most reason is the costs of matching.
Dieter know this too. Or isn't it ???
And this would be one way (perhaps Nelson Pass's way) to solve the R66-Jumper problem. But an expensive way. Which DIY'er buys hundrets of FETs to match out 8 ?
It's easier to calculate the 30dB gain with new values for RS and R40 following for high gm (>22mS), low drain current (< 2,5 mA/per FET) and a good operating point which should not drop under 15V at the collector of the cascode Q14.
For those which want to match their FETs and use the original values :
The 2SK170s should be matched to a good value of gm ( > 22mS), by matching the FETs to same IDss with very low gate pitch-off voltage "Vp".
In other words : try to match the FETs with 22 Ohm source resistor to max. 2,5mA drain current and look for gm >22mS.
Because gm is also a function of IDss and Vp, theoretical all IDss classification types ( GR, BL, V ) can be used. I thing it's better to use BL or V types, which have principal higher IDss values.
With a lower pitch-off voltage, you can achieve lower drain currents by using the same value of RS.
For those who understand the theorie, the drain curve in the tranconductance figure of the datasheet plunge down more from IDss before it rounds up than with higher Vp-values (square law form of the drain curve).
All others which had build their Ono and want to use the MC-Input and have measured that the sum of drain current through all Input-FETs (Q10-Q13) is greater than 10mA : Install the jumper over R66. The MC-Preamp will function. The only disadvantage is, that the amp produces more current noise.
@hsort
I use Orcad 9.2.2. which includes Pspice. The 2SK170 is part of the japan JFET library. But i've build own models with different IDss values.
@Dieter
Gibts bei dir au Prozentle ?
toshiba
Hi LSV-Dieter, nice to meet you here!
Don't look at possible literal mistakes in the schematics of the Ono! Look at the circuit and try to understand it.
If the output-voltage of the MC-sub-regulator is really 26 Volt,
the cascode Q14 will be biased at it's base with about 13V.
So the Drain voltage can't be 15V !
Another (possible) mistake is "SHORT for -4db". Halving the gain-resistors (R66+R40) of the first stage normally gives -6dB. (Ok, I don't look at the following load-resistance)
And if you take a look here and see what Nelson Pass prints on his boards : "Remove for high gain + 3dB"
What is true ???
hsort, why to increase the rail, only why there is mark "15 VOLTS". I think 26 Volt is more than enough to get the amp work.
Yes, to decrease the idle current by increasing Rs is one way.
But what's about the gain (30dB) and those one, whose output of the MC-System is equal or less than 100uV. Good input value to the MM is about 10 mV.
You can calculate the gain in a simply way :
4 * (R66 + R40) / (RS + 1/gm)
"RS" is value of the source resistor (22) and "gm" the transconductance of the FETs (with assumption that all FETs have the same gm). Look at the datasheet of the 2SK170. Typical value of gm is 22mS. With R66 shorted, you get about 30 dB.
Sorry I lost the way !
Why are Pass amps so expensive and the circuits so simple ?
I think one of the most reason is the costs of matching.
Dieter know this too. Or isn't it ???
And this would be one way (perhaps Nelson Pass's way) to solve the R66-Jumper problem. But an expensive way. Which DIY'er buys hundrets of FETs to match out 8 ?
It's easier to calculate the 30dB gain with new values for RS and R40 following for high gm (>22mS), low drain current (< 2,5 mA/per FET) and a good operating point which should not drop under 15V at the collector of the cascode Q14.
For those which want to match their FETs and use the original values :
The 2SK170s should be matched to a good value of gm ( > 22mS), by matching the FETs to same IDss with very low gate pitch-off voltage "Vp".
In other words : try to match the FETs with 22 Ohm source resistor to max. 2,5mA drain current and look for gm >22mS.
Because gm is also a function of IDss and Vp, theoretical all IDss classification types ( GR, BL, V ) can be used. I thing it's better to use BL or V types, which have principal higher IDss values.
With a lower pitch-off voltage, you can achieve lower drain currents by using the same value of RS.
For those who understand the theorie, the drain curve in the tranconductance figure of the datasheet plunge down more from IDss before it rounds up than with higher Vp-values (square law form of the drain curve).
All others which had build their Ono and want to use the MC-Input and have measured that the sum of drain current through all Input-FETs (Q10-Q13) is greater than 10mA : Install the jumper over R66. The MC-Preamp will function. The only disadvantage is, that the amp produces more current noise.
@hsort
I use Orcad 9.2.2. which includes Pspice. The 2SK170 is part of the japan JFET library. But i've build own models with different IDss values.
@Dieter
Gibts bei dir au Prozentle ?
toshiba
>>@Dieter
>>Gibts bei dir au Prozentle ?
@toshiba
thanks for the detailed description. I use the MC-amp with a Ortofon MC2 and the jumper installed.
Noise is absolutely no problem.
Prozentle auf was?
Einfach mal anrufen. Bin am besten Abends zu erreichen.
Tel.-Nr auf der Homepage.
Regards
Dieter
>>Gibts bei dir au Prozentle ?
@toshiba
thanks for the detailed description. I use the MC-amp with a Ortofon MC2 and the jumper installed.
Noise is absolutely no problem.
Prozentle auf was?
Einfach mal anrufen. Bin am besten Abends zu erreichen.
Tel.-Nr auf der Homepage.
Regards
Dieter
Hi Toshiba,
i agree with your statements , but i am wondering the given items in the inputstage of the ono.the divider for the cascode gives about 13 V at the base , so the emitter gives about 12,4 V and not 15 V in the circuit.You get only 15 V with a higher rail.That is , what i pointed out.
And in the easier pearl version , you can see , that the loadresistor is 499 here with 2 sk 170 bl.this give about 20 ma iddle current.
orcad 9.2 is a full version or is it free , but reduced version ?
where can i get this version ?
i agree with your statements , but i am wondering the given items in the inputstage of the ono.the divider for the cascode gives about 13 V at the base , so the emitter gives about 12,4 V and not 15 V in the circuit.You get only 15 V with a higher rail.That is , what i pointed out.
And in the easier pearl version , you can see , that the loadresistor is 499 here with 2 sk 170 bl.this give about 20 ma iddle current.
orcad 9.2 is a full version or is it free , but reduced version ?
where can i get this version ?
Hi hsort,
i agree with you too. If emitter should be 15V the rail must be higher.
Correction : in my first statement of this thread, i talk about Ub=15V of Q14. It's wrong, because i calculate with 30V PSU not with 26V rail-output of the MC sub regulator.
Orcad : There exists full version ( expensive ) and evaluation versions which are free.
Look here
toshiba
i agree with you too. If emitter should be 15V the rail must be higher.
Correction : in my first statement of this thread, i talk about Ub=15V of Q14. It's wrong, because i calculate with 30V PSU not with 26V rail-output of the MC sub regulator.
Orcad : There exists full version ( expensive ) and evaluation versions which are free.
Look here
toshiba
Hi all,
I`m now almost ready getting all the parts for my Ono and did some 2SK170BL matching tonight.
I measured Vp and Idss for the 25 parts and could find two sets of four wich match quite reasonably. The Problem is that these all have an Idss around 6.4mA and Vp around -0.62V.
With a 22 ohms source resistor this ´ll give around 4.2mA per Fet wich is a bit much as I read earlier in this thread.
To get the current down to 2.5mA per Fet the source resistors must be 100 ohms. That isn`t the problem......
The problem is that this will also reduce the gain to 27x or 29dB.
This is almost the same value wich I would get when I would leave the source resistor at 22 Ohm and short R40 or R66)
So does this mean that for this stage to work with maximum gain I would need 2sk170 with lower Idss???
The second question is what would theoretically be better:
1. 29dB gain with 4.2mA per Fet (R40 shorted), Rs 22 Ohm
2. 29dB gain with 2.5mA per Fet Rs 100 Ohm
william
I`m now almost ready getting all the parts for my Ono and did some 2SK170BL matching tonight.
I measured Vp and Idss for the 25 parts and could find two sets of four wich match quite reasonably. The Problem is that these all have an Idss around 6.4mA and Vp around -0.62V.
With a 22 ohms source resistor this ´ll give around 4.2mA per Fet wich is a bit much as I read earlier in this thread.
To get the current down to 2.5mA per Fet the source resistors must be 100 ohms. That isn`t the problem......
The problem is that this will also reduce the gain to 27x or 29dB.
This is almost the same value wich I would get when I would leave the source resistor at 22 Ohm and short R40 or R66)
So does this mean that for this stage to work with maximum gain I would need 2sk170 with lower Idss???
The second question is what would theoretically be better:
1. 29dB gain with 4.2mA per Fet (R40 shorted), Rs 22 Ohm
2. 29dB gain with 2.5mA per Fet Rs 100 Ohm
william
Hi wuffwaff,
first i want to address Mr. Nelson Pass and Mr. Wayne Colburn who might give us a statement about best idle current of the ONO MC-stage.
To increase the gain (RS=100), increase the value of R40 or R66.
But don't forget to observe the collector potential of Q14. IMHO it should not drop under Vbase(Q14) + 2V.
To your first question :
It's not necessary to find Fet's with lower Idss. The only thing to do is to play with the R-values of the gain formula : 4 * (R66 + R40) / (RS + 1/gm). But once again. Look for good working conditions of cascode Q14.
To your second question :
Do a hearing session and find out which one gives better results.
As i wrote in another thread, for the session let warm up the ONO.
toshiba
first i want to address Mr. Nelson Pass and Mr. Wayne Colburn who might give us a statement about best idle current of the ONO MC-stage.
To increase the gain (RS=100), increase the value of R40 or R66.
But don't forget to observe the collector potential of Q14. IMHO it should not drop under Vbase(Q14) + 2V.
To your first question :
It's not necessary to find Fet's with lower Idss. The only thing to do is to play with the R-values of the gain formula : 4 * (R66 + R40) / (RS + 1/gm). But once again. Look for good working conditions of cascode Q14.
To your second question :
Do a hearing session and find out which one gives better results.
As i wrote in another thread, for the session let warm up the ONO.
toshiba
Hello Toshiba,
well that is the problem.
If I make the Rs´s higher I`ll have to make the (R40 + R67) higher too. (to have the same gain).
This will make the voltage drop over R40+R67 again too high for the cascode to work (at 35 dB gain).
So that is why I thought that the only way to get the cascode working at the higher gain setting is to lower the current without raising the source resistors (Lower Idss)
For the second question I only would like to know what will theoretically give the lowest noise:
2.5 mA through a 100R resistor or
4.2 mA through a 22R resistor
A hearing session is not possible at the moment cause I´m still waiting for some parts (transformers etc.) cause I had some trouble at the place I ordered them and now have to find new suppliers
Brownlow,
I also got the pquadrat boards but things tend to take a long time when you´ve got too many hobbys So I´m afraid this is not a new chance...
william
well that is the problem.
If I make the Rs´s higher I`ll have to make the (R40 + R67) higher too. (to have the same gain).
This will make the voltage drop over R40+R67 again too high for the cascode to work (at 35 dB gain).
So that is why I thought that the only way to get the cascode working at the higher gain setting is to lower the current without raising the source resistors (Lower Idss)
For the second question I only would like to know what will theoretically give the lowest noise:
2.5 mA through a 100R resistor or
4.2 mA through a 22R resistor
A hearing session is not possible at the moment cause I´m still waiting for some parts (transformers etc.) cause I had some trouble at the place I ordered them and now have to find new suppliers
Brownlow,
I also got the pquadrat boards but things tend to take a long time when you´ve got too many hobbys
So I´m afraid this is not a new chance...william
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.