Well, ive put the output boards together pretty much and i thought id try running a signal through the riaa network.
Within the accuracy of my scope they are perfectly matched. (i know this isnt quite the case but itll have to do).
Its not quite conforming to the riaa curve. I suspect that once the tube is in place thatll affect it (hopefully improve it).
Having fed the signal in directely to the tube end of R7 with no tube attched I got a drop of ~18.2dB between 20hz and 1khz. My scope is connected across the 1M resistor R10.
I did try putting 100k across R7 and 2k2 in front of R7 to see if that increased the ratio between 50hz and 1khz. lowering the value of R7 seems to reduce the ratio. I just tried sticked 2k2 in front and that has edged the ratio to 18.98dB so once its all up and running and can test start to finish Ill maybe increase R7 to see if i can get the required 19.74dB (taken from here RIAA Equalization).
Something that I am a bit concerned with though is although the circuit handles sine waves wonderfully, seemingly to conform to riaa etc. and starting to tail off at 7Hz, square wave performance here is abysmal(not sure this is the right choice of word). I would expect the low freq square waves to sharply slope after the transient due to the blocking caps but it seems that at pretty much any frequency i get a rather distorted triangle wave. Presumably this is also due to very HF rolloff removing the high harmonics that would make a squarewave actually square.
I guess a record cant produce a square wave like a cd can (or my synthesizers which produce awesome square waves!) but yet instruments that produce square waves seem to sound fine played back on either format.
What am i missing? Or is this just fine.
Its also making me requestion (again) how i go about selecting the best loading for my input transformers.
Within the accuracy of my scope they are perfectly matched. (i know this isnt quite the case but itll have to do).
Its not quite conforming to the riaa curve. I suspect that once the tube is in place thatll affect it (hopefully improve it).
Having fed the signal in directely to the tube end of R7 with no tube attched I got a drop of ~18.2dB between 20hz and 1khz. My scope is connected across the 1M resistor R10.
I did try putting 100k across R7 and 2k2 in front of R7 to see if that increased the ratio between 50hz and 1khz. lowering the value of R7 seems to reduce the ratio. I just tried sticked 2k2 in front and that has edged the ratio to 18.98dB so once its all up and running and can test start to finish Ill maybe increase R7 to see if i can get the required 19.74dB (taken from here RIAA Equalization).
Something that I am a bit concerned with though is although the circuit handles sine waves wonderfully, seemingly to conform to riaa etc. and starting to tail off at 7Hz, square wave performance here is abysmal(not sure this is the right choice of word). I would expect the low freq square waves to sharply slope after the transient due to the blocking caps but it seems that at pretty much any frequency i get a rather distorted triangle wave. Presumably this is also due to very HF rolloff removing the high harmonics that would make a squarewave actually square.
I guess a record cant produce a square wave like a cd can (or my synthesizers which produce awesome square waves!) but yet instruments that produce square waves seem to sound fine played back on either format.
What am i missing? Or is this just fine.
Its also making me requestion (again) how i go about selecting the best loading for my input transformers.
Ive tried putting the tubes in but that doesnt seem to be affecting it too much. However,
from the article....
"Trimming the RIAA network isn't too hard. Adjust the buildout resistor R7 to get the 1kHz response to be 20 dB below the 50Hz response. Then adjust C4 to give response at 20kHz at about -19dB with respect to 1kHz (if you're using the Missing Zero) or -19.6dB if you're not."
1. Im finding that lowering R7 (20k) reduces the ratio between 50hz and 1khz not increasing it.
2. According to the RIAA db chart ive found I think that the 50hz to 1khz should only be 16.95db, 10hz to 1khz should be 19.74 (close enough to 20db).
3. Reducing my R8 (I used a 3.15k resistor) with a 15k in parallel sorts the ratio out almost exactly.
4. Adjust C4 to get between 1khz and 20khz. Shouldnt that be C2?
So,
Sticking 15k across R8 (giving ~2.6K) and an extra 3n3 on to C4
gives me the following
Im not using the 91R resistor, thats just shorted but that seems to be giving me the response as if i were.
I really dont think i could hear 0.5 db at 20khz so i think thats probably close enough.
from the article....
"Trimming the RIAA network isn't too hard. Adjust the buildout resistor R7 to get the 1kHz response to be 20 dB below the 50Hz response. Then adjust C4 to give response at 20kHz at about -19dB with respect to 1kHz (if you're using the Missing Zero) or -19.6dB if you're not."
1. Im finding that lowering R7 (20k) reduces the ratio between 50hz and 1khz not increasing it.
2. According to the RIAA db chart ive found I think that the 50hz to 1khz should only be 16.95db, 10hz to 1khz should be 19.74 (close enough to 20db).
3. Reducing my R8 (I used a 3.15k resistor) with a 15k in parallel sorts the ratio out almost exactly.
4. Adjust C4 to get between 1khz and 20khz. Shouldnt that be C2?
So,
Sticking 15k across R8 (giving ~2.6K) and an extra 3n3 on to C4
gives me the following
Im not using the 91R resistor, thats just shorted but that seems to be giving me the response as if i were.
I really dont think i could hear 0.5 db at 20khz so i think thats probably close enough.
Last edited:
Just set all my css.
d3a
19.77mA 19.67mA
6922a
9.54mA 9.57mA
6922b
9.52mA 9.56mA
Ive just been having a look through this topic and unfortunately couldnt find it but i remember that someone had set their css quite a bit lower, something like 16ma for the d3a and 6ma or 8 ma for the 6922.
Ive just looked at the datasheet for the 6922eh and it looks like its saying the plate current should be 10mA min to 20mA max. Have i missed something here?
Should i set the current a bit higher? Mine are all coming in at ~ 9.5mA or do you think thats ok?
As i hadd found them here with minimal tweaking i thought slightly lower would be better than slightly higher.
Same goes for the D3a, i think thats meant to be 22mA with 20mA being the lowest.
The article says that exactitude isnt needed but I assumed that 10mA and 20mA werent on the lower limit of the devices.
Thanks
d3a
19.77mA 19.67mA
6922a
9.54mA 9.57mA
6922b
9.52mA 9.56mA
Ive just been having a look through this topic and unfortunately couldnt find it but i remember that someone had set their css quite a bit lower, something like 16ma for the d3a and 6ma or 8 ma for the 6922.
Ive just looked at the datasheet for the 6922eh and it looks like its saying the plate current should be 10mA min to 20mA max. Have i missed something here?
Should i set the current a bit higher? Mine are all coming in at ~ 9.5mA or do you think thats ok?
As i hadd found them here with minimal tweaking i thought slightly lower would be better than slightly higher.
Same goes for the D3a, i think thats meant to be 22mA with 20mA being the lowest.
The article says that exactitude isnt needed but I assumed that 10mA and 20mA werent on the lower limit of the devices.
Thanks
I can't speak to the Master's Noise or the D3a, but in the Equal Opportunity article SY references an article that suggests lower noise for the 6922 series at lower current.
"Figure 11 shows the input gain block and RIAA equalization network (the odd numbering will be explained
later). Since the first part appeared, I read Blencowe’s first-rate and badly needed article on
tube noise1, and was fascinated to find that the optimum current for low noise in the ECC88 family was
4mA. This is much lower than folk wisdom suggests. More current, more gm, lower noise, let’s go
home! I’m guilty of that, and have accordingly lowered the gain block’s idle current to 4mA. Now, the
rise in noise with increasing current is shallow, so I can weakly defend all of my earlier choices of high
current operating conditions for input stage tubes by claiming that they weren’t much worse than
the optimum, and only cost more heat and somewhat reduced tube life. What that also means is that
you needn’t obsess on hitting 4mA exactly, there’s a broad plateau in the noise versus current curves.
The other advantage of the 4mA idle current is that your choice of pFETs within a lot will be broader
since the minimum idss for the LSJ74 B series (the only series I have been able to source) is 6mA.
Linear Audio Vol 8 - September 2014 71
1 Merlin"
When I look at ECC88 and 6DJ8 datasheets, I find 15 mA typical plate current, but no minimum. The curves suggest 6 or 8 mA would make a fine operating point for a preamp.
I don't know if that's the reason why someone went to lower current, but it doesn't look like a bad idea. It would also improve tube life by reducing stress.
Jac
"Figure 11 shows the input gain block and RIAA equalization network (the odd numbering will be explained
later). Since the first part appeared, I read Blencowe’s first-rate and badly needed article on
tube noise1, and was fascinated to find that the optimum current for low noise in the ECC88 family was
4mA. This is much lower than folk wisdom suggests. More current, more gm, lower noise, let’s go
home! I’m guilty of that, and have accordingly lowered the gain block’s idle current to 4mA. Now, the
rise in noise with increasing current is shallow, so I can weakly defend all of my earlier choices of high
current operating conditions for input stage tubes by claiming that they weren’t much worse than
the optimum, and only cost more heat and somewhat reduced tube life. What that also means is that
you needn’t obsess on hitting 4mA exactly, there’s a broad plateau in the noise versus current curves.
The other advantage of the 4mA idle current is that your choice of pFETs within a lot will be broader
since the minimum idss for the LSJ74 B series (the only series I have been able to source) is 6mA.
Linear Audio Vol 8 - September 2014 71
1 Merlin"
When I look at ECC88 and 6DJ8 datasheets, I find 15 mA typical plate current, but no minimum. The curves suggest 6 or 8 mA would make a fine operating point for a preamp.
I don't know if that's the reason why someone went to lower current, but it doesn't look like a bad idea. It would also improve tube life by reducing stress.
Jac
Thanks jac, well, it's a 6922eh that I'll be using. The Datasheet I found doesn't look particularly comprehensive. But at least I know other tubes are happy with less than 10ma so I should be safe. Cases didn't arrive today but everythings ready to go. I may experiment with lowering the current later.
Hi Frank. Am wondering the reasoning for sy specifying 20ma and10ma. I'm assuming it's where the tubes are most linear and/or give the most headroom. My cartridge is higher output than the test example at 0.3 for 3.54 (?) cms a second so defo need to not reduce headroom. What's the output on yours? Cheers.
The replacement for the first stage is a 6J52P (not 6Z52).
I found it by searching for a tube, cause i´m building a Hybrid Amp with Tube as first stage. There is a D3A used, but for my personal needs is the gain to high. Full power at 400mV.
Please don´t understand wrong. The 6J52P has a similar gain like the D3A. I´m now using a 6C45 for the Hybrid.
I think, in a few years, there will be no more D3A at the marketplace. So it´s becalming to have a replacement tube.
I found it by searching for a tube, cause i´m building a Hybrid Amp with Tube as first stage. There is a D3A used, but for my personal needs is the gain to high. Full power at 400mV.
Please don´t understand wrong. The 6J52P has a similar gain like the D3A. I´m now using a 6C45 for the Hybrid.
I think, in a few years, there will be no more D3A at the marketplace. So it´s becalming to have a replacement tube.
brilliant, thanks guys.
I am going to leave it at the stock current. One of the reasons is that Ive just tested my psu and the raw voltage rectified and unloaded is coming out at 400v. I think i need to keep the load higher rather than lower.
The guy who designed the transformer for me (it is a 250-0-250) used a low flux (i think) design and said it will come out a bit on the hot side but will sag quite a bit under load to come out where i need it. I told him that it would have 80mA load on it.
The voltage does take a few seconds to stabilise at 400v (unloaded). Im hoping once a load is connected it will be a little slower and it should come out at a lower voltage than that too.
I think that both types of regs will draw 20mA each
I think the voltage drop should be 240V max on the 160v reg. It should be less and Id imagine there would be quite a drop on the 1k8 5w resistor. If 20mA is going through then i make that 36v
Therefore 204v across the tip 50. 204v x 0.02A = 4.08W max.
For the 260v reg that'll drop 140v across the tip 50 so 140v x 0.02A 2.8W.
I have put reasonably big heatsinks on all devices but at 12.9c/w then 4 watts is going to push 52C on top of ambient which inside the case is too hot.
It looks like the supply needs to sag about 300v rectified to be comfortably safe.
Its a 50VA transformer. Anything i can do to try this safely? I have a light bulb tester.
My PEM has 2 fuses rather annoyingly. So i put a 2amp on the neutral line and i put a 500mA (smallest i have) on the live so the live ruptures first.
Any suggestions?
Cheers
I am going to leave it at the stock current. One of the reasons is that Ive just tested my psu and the raw voltage rectified and unloaded is coming out at 400v. I think i need to keep the load higher rather than lower.
The guy who designed the transformer for me (it is a 250-0-250) used a low flux (i think) design and said it will come out a bit on the hot side but will sag quite a bit under load to come out where i need it. I told him that it would have 80mA load on it.
The voltage does take a few seconds to stabilise at 400v (unloaded). Im hoping once a load is connected it will be a little slower and it should come out at a lower voltage than that too.
I think that both types of regs will draw 20mA each
I think the voltage drop should be 240V max on the 160v reg. It should be less and Id imagine there would be quite a drop on the 1k8 5w resistor. If 20mA is going through then i make that 36v
Therefore 204v across the tip 50. 204v x 0.02A = 4.08W max.
For the 260v reg that'll drop 140v across the tip 50 so 140v x 0.02A 2.8W.
I have put reasonably big heatsinks on all devices but at 12.9c/w then 4 watts is going to push 52C on top of ambient which inside the case is too hot.
It looks like the supply needs to sag about 300v rectified to be comfortably safe.
Its a 50VA transformer. Anything i can do to try this safely? I have a light bulb tester.
My PEM has 2 fuses rather annoyingly. So i put a 2amp on the neutral line and i put a 500mA (smallest i have) on the live so the live ruptures first.
Any suggestions?
Cheers
Hi mrdave,
first test how much the raw voltage under load will be. Use a 10K Resistor as load (400V/40mA) for testing. Now measure the raw voltage.
For a good filtering are 50V above the output voltage needed. This is enough.
You can drop your raw voltage to 310V with an ordinary resistor. I used Vishay Dale with metalbody mount at an well ventilated point of the case.
first test how much the raw voltage under load will be. Use a 10K Resistor as load (400V/40mA) for testing. Now measure the raw voltage.
For a good filtering are 50V above the output voltage needed. This is enough.
You can drop your raw voltage to 310V with an ordinary resistor. I used Vishay Dale with metalbody mount at an well ventilated point of the case.
I built the phono stage for my Hovland HP 100. There is no Phono-Stage inside, but the PS of the Hovland is build for an additional Phono Stage. So i got 415V from the existing PS.
I dropped the voltage to 315V with 2.5K x 40mA for both regulators and both channels. From 315V to the 160V-regulator i used one more additional Resistor (Mills 7,5k). So i got 240V as input-voltage. Therefor i lost the 1,8k Resistor.
My TIP50 are only hand-hot. i used no heatsinks, only the metalbody of the case.
The solution without heatsinks use only with the lower current, please.
But with the additional resistors your heatsinks will become not too hot (1W for every heatsink, if the input voltage will be 50V).
I dropped the voltage to 315V with 2.5K x 40mA for both regulators and both channels. From 315V to the 160V-regulator i used one more additional Resistor (Mills 7,5k). So i got 240V as input-voltage. Therefor i lost the 1,8k Resistor.
My TIP50 are only hand-hot. i used no heatsinks, only the metalbody of the case.
The solution without heatsinks use only with the lower current, please.
But with the additional resistors your heatsinks will become not too hot (1W for every heatsink, if the input voltage will be 50V).
Thanks, that's given me some useful clues how to fix this. As I have 4 ht regs all having 20ma draw I like the idea of sticking a 2k5 resistor on the front of each. That means that the resistor is only required to disipate 1w whilst giving a 50v drop. Worst case that means that I would have 400-50-36-160 across tip50 which is ~3w. If I were to stick a second 2k5 there then it would be nearer 2w on tip50.
I think I prefer this approach than a big power resistor on the main ht.
Regarding testing the sag, a 10k resistor will need to be able disipate 16w at 400v. hmm. I dont think I have anything like that kicking about.
I think I prefer this approach than a big power resistor on the main ht.
Regarding testing the sag, a 10k resistor will need to be able disipate 16w at 400v. hmm. I dont think I have anything like that kicking about.