pagan
Been here a while!
Posts: 512
|
Post by pagan on Feb 23, 2013 8:40:03 GMT
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 23, 2013 10:28:31 GMT
Allan You need to be mindful of a couple of things here.We could have used 2,200uF 6.3V electros, but electros with a much higher ripple rating are desirable for reliability, especially if used with an SMPS supply. Are you able to obtain the BC8xx series in "c"versions ? Are YOU able to check their HFE ? Alex
|
|
Will
Been here a while!
Ribena abuser!
Member since 2008
Posts: 2,164
|
Post by Will on Feb 23, 2013 11:18:27 GMT
Hi Allan,
I've been thinking a SMT JLHRE for a bit as well. For the toshies, I was thinking of SMT versions of the BC639/40, as used in older versions of the RE. It limits things a bit, but perhaps with larger pcb 'sinks' 100mA in a CL could be done?
I did think of using normal though hole caps for the 2200uf, 4700uf, but perhaps there are SMT versions of those rubycon caps Alex is using?
|
|
pagan
Been here a while!
Posts: 512
|
Post by pagan on Feb 23, 2013 11:59:07 GMT
For the Toshiba's Just quick search thought RS looking for ~2 Watt devices Found these Sanyo VC2102 australia.rs-online.com/web/p/bipolar-transistors/0546622/Package still smt Doesn't really matter the package as long as the devices can handle power, and the parameters are up to it.. Up to 2 amp, prefer 50V but these are 30V high gain, good fT, cob could be better but this is early searches. Alex BC846 B's are there, will search for others. caps australia.rs-online.com/web/c/passive-components/capacitors/aluminium-capacitors/?sort-by=default&sort-order=default&applied-dimensions=4294857513,%204294789462,%204294518889,%204294516023,%204294509049,%204294518167,%204294519046,%204294516375,%204294518998,%204294518156,%204294519035,4294881131,4294845545,4294829570,4294832604,4294832601&lastAttributeSelectedBlock=4294958098
|
|
pagan
Been here a while!
Posts: 512
|
Post by pagan on Feb 23, 2013 13:04:28 GMT
Hi Allan, I've been thinking a SMT JLHRE for a bit as well. For the toshies, I was thinking of SMT versions of the BC639/40, as used in older versions of the RE. It limits things a bit, but perhaps with larger pcb 'sinks' 100mA in a CL could be done? I did think of using normal though hole caps for the 2200uf, 4700uf, but perhaps there are SMT versions of those rubycon caps Alex is using? Hi Will There are a good range of SMT trannies around now, even in dual packages. For Toshies alternatives, we need the power capabilities first then go from there. For the BC's ... high hfe, 30v to 50v low cob, preferable to have npn and pnp versions. the search continues. Allan
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 23, 2013 15:49:38 GMT
Happy hunting Allan and Will, I'm watching closely
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 1, 2013 23:42:25 GMT
Hi Allan,
I have had similar thoughts regarding a SMT JLH ripple eater but..............
I would recommend testing any new caps in the existing PCB before to much effort is put into a new SMT board. As you know, these caps can be show stoppers.
I have tried 2 x 1800uF 2.5V OSCONS and was disappointed with the results. Now I did accidently connect these caps up to 15VAC, so they may have been damaged. Also, disappointing were 1 x 8200uF 6.3V caps and I have even tried 2 x 100uF caps and they were no good.
On the other hand, I have tried most of the 2200uF and 4700uF Suntans with good results, though I can't hear differences between 2200uF and 4700uF.
BTW: If anyone would like to send me caps to test, I will be able to test them in my SCDAC. I have plenty of JLH ripple eaters and loads of time.
regards
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 2, 2013 0:09:30 GMT
Hi Greg With a direct coupled preamp/HA/PA especially, the lows are further improved. In the case of the 15W Class A you can actually feel LF more with suitable material and good speakers, I think that Miguel liked the small gains in that area with the SC (Jaycar) HA too after doing that. You can't presently do that with the CL section in use though.
Kind Regards Alex
P.S. Did you see my comments about trying the heavy wire link between In and Out of the Class A preamp's JLHs earth terminals? It sounded a little like the capacitance multiplier section was working, but the ripple detector section was degraded due to the bypassing .
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 2, 2013 0:55:06 GMT
Yes Alex,
I am only testing the JLH ripple eater in my SCDAC at the moment. I was just pointing out, that even I can hear differences with some caps on the JLH ripple eater, so it is a worthwhile experimenting with the SMT caps before putting too much effort into a SMT PCB.
Hopefully, the newer JLH ripple eater PCB (GE20120421) should address both issues as it has a bigger, thicker earth trace. I must convert one of my JLH ripple eaters to the new PCB for comparison.
regards
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 2, 2013 6:41:06 GMT
Hi Greg Sorry to be a wet blanket, but I doubt that the small SMD electros have anywhere near the ripple ratings required for longevity. Kind Regards Alex P.S. I also prefer Vitabrits over Weetbix, but I will buy Weetbix if it's on special !
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 6, 2013 8:31:56 GMT
Hi guys,
I would like a deeper understanding of the JLH Ripple Eater and how it works, so I can make a few improvements (if necessary) on the next PCB.
I'd like to start at the Constant Current Source (CCS). We often refer to this as the Current Limiter, but I think that is inaccurate. The circuit at the front of the ripple eater (D2, D3 R12, Q6 and R11) forces a constant current into the ripple eater and load. A current limiter to me would allow a variable current from near 0 to a maximum setting. From what I understand, Iconst is equal to the voltage across D2, D3 (2 x 0.7v) minus Q6 Vbe (0.7) divided by R12. But what is the function of R11. I thought R11 sets the minimum current through D2, D3 string and/or ensures Q6 is switched on. Help??
Looking at other regulator circuits, the CCS is often a feature of the regulator not an option. Reading the JLH article it sort of infers it is only for protecting the circuit by restricting the current during the initial charging of C4. What do you think? Should we be always using it?
Are there any LTSpice experts out there? The more simulations I do.. only makes me realise how my understanding is limited.
BTW: 1.3 kg Weetbix are regularly on special for $5 at Coles.
regards Greg
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 6, 2013 9:10:43 GMT
Hi guys, I would like a deeper understanding of the JLH Ripple Eater and how it works, so I can make a few improvements (if necessary) on the next PCB. I'd like to start at the Constant Current Source (CCS). We often refer to this as the Current Limiter, but I think that is inaccurate. The circuit at the front of the ripple eater (D2, D3 R12, Q6 and R11) forces a constant current into the ripple eater and load. A current limiter to me would allow a variable current from near 0 to a maximum setting. From what I understand, Iconst is equal to the voltage across D2, D3 (2 x 0.7v) minus Q6 Vbe (0.7) divided by R12. But what is the function of R11. I thought R11 sets the minimum current through D2, D3 string and/or ensures Q6 is switched on. Help?? Looking at other regulator circuits, the CCS is often a feature of the regulator not an option. Reading the JLH article it sort of infers it is only for protecting the circuit by restricting the current during the initial charging of C4. What do you think? Should we be always using it? Are there any LTSpice experts out there? The more simulations I do.. only makes me realise how my understanding is limited. BTW: 1.3 kg Weetbix are regularly on special for $5 at Coles. regards Greg Hi Greg The Constant Current Source is at the output end of the JLH. It governs the amount of current available for modulating purposes via the 33 ohm emitter resistor.( typically around 16mA at +-15V supply rails ) However, this available current can vary in value depending on the supply rail voltage, which is why at voltages as low as 5V and 3.3V a 1.4mA (approx.) J506 Current Regulating diode is used in place of the 10K resistor for best results. This keeps the standby current fairly close in value to that at 15V with the 10K resistor. The Current Limiter section at the supply input side merely prevents the amount of current drawn at start up from exceeding preset limits governed by the value of it's emitter resistor etc. You can think of it as being a simple series DC switch until the current drawn approaches the set limits as per the formula in the diagrams posted. With no load at the output.it will only pass a little more than the current set by the CCS through the 33 ohm resistor. Actually, the original CL section is very basic, and has severe limitations at currents more than the 100mA maximum shown in the chart by J.L.H. I suspect that this is the reason that the original schematic only showed suitable values for up to 100mA instead of the 300mA mentioned elsewhere. If possible, it is far better to use the more sophisticated Current Limiting and over temperature sections built into a preceding voltage regulator. R11 provides a suitable bias current to turn the CL section transistor on as a switch , and at currents well below the CL setting the transistor 's voltage drop between emitter and collector will be at close to the saturation voltage as given in the data sheets. As we are using the J.L.H at higher currents than the original design, the base current had to be increased in order to keep this voltage drop low. Kind Regards Alex
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 6, 2013 11:59:21 GMT
Thanks Alex,
You are right of course!! Just after the previous post I found a spice tutorial on this current regulator circuit and I can now see how it works. I think I know how to tune it now for lower voltages... danger!!
regards Greg
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 6, 2013 22:14:15 GMT
Thanks Alex, You are right of course!! Just after the previous post I found a spice tutorial on this current regulator circuit and I can now see how it works. I think I know how to tune it now for lower voltages... danger!! regards Greg Hi Greg It's not quite that easy. For example, you could reduce the 10K resistor to 3.3K and get a fairly similar standby current as at 15V, but the CRD works better , most likely because it is a constant current source. Many years ago with the LH0002 buffer I.C. I made up a discrete version, and replaced the published schematic value collector resistors (5.1K) with FETs connected as CRDs and the buffers sounded better ! They were used that way in an earlier preamp that you may have even heard years ago.(Geoff has it now) Kind Regards Alex P.S. I mentioned this to Silicon Chip, and several months later they published a video switcher using their own version of a discrete LH0002.buffer .
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 0:03:37 GMT
Hi Alex,
Thanks for information.
I was going to suggest 3 things for the CL.
One, maybe replace the diode string D2, D3 with a LED as is the current trend. This simplifies the circuit/PCB as well as given a visual queue that something is happening. There are other options for a voltage reference but we still want something simple.
Two, (as you suggested) replace R11 with a constant current source (CCS). It seems common at the moment to use a FET (or two) connect as a CCS is a popular option. The problem I see with the JLH Ripple Eater generally is each section with not work as effectively when the voltages are changed from the original 15V.
Three, a cap on the input and/or across the voltage reference. My JLH Ripple Eater are over 10cm from the PSU.
I think, CL and CCS circuits present a much higher resistance than their effective resistance. The original CL has a output resistance in the order of 500K. This output resistance reduces when the CL is tuned for 5 volts. Should we be aiming for a higher or lower output resistance?? Usually higher?? I looks like the smaller the voltage the greater the need for R11 to be replaced with a CCS to get the output resistance back up??
I have been thinking, when we remove the CL we are also removing this output resistance so the JLH Ripple Eater now sees the output resistance of the PSU instead. Is the CL also acting as a buffer?
regards Greg
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 0:57:36 GMT
Hi Alex, Thanks for information. I was going to suggest 3 things for the CL. One, maybe replace the diode string D2, D3 with a LED as is the current trend. This simplifies the circuit/PCB as well as given a visual queue that something is happening. There are other options for a voltage reference but we still want something simple. Two, (as you suggested) replace R11 with a constant current source (CCS). It seems common at the moment to use a FET (or two) connect as a CCS is a popular option. The problem I see with the JLH Ripple Eater generally is each section with not work as effectively when the voltages are changed from the original 15V. Three, a cap on the input and/or across the voltage reference. My JLH Ripple Eater are over 10cm from the PSU. I think, CL and CCS circuits present a much higher resistance than their effective resistance. The original CL has a output resistance in the order of 500K. This output resistance reduces when the CL is tuned for 5 volts. Should we be aiming for a higher or lower output resistance?? Usually higher?? I looks like the smaller the voltage the greater the need for R11 to be replaced with a CCS to get the output resistance back up?? I have been thinking, when we remove the CL we are also removing this output resistance so the JLH Ripple Eater now sees the output resistance of the PSU instead. Is the CL also acting as a buffer? regards Greg Hi Greg Just a couple of quick thoughts. Most LEDS these days have a forward voltage of > 1.8V. We would need to take that into account. LEDs also have a fairly low reverse voltage rating. We would need somebody like Marc perhaps, to think about any possible pitfalls. My experience is that above 12V there is no audible advantage to replacing the 10K with a far more expensive CRD. Above 24V or so it may be worthwhile increasing the value of the 10K resistor in proportion to the increase of the supply voltage. When using the JLH at higher currents than the original design, there is a definite SQ advantage in going to parallel 4,700uF capacitors instead of the original 2,200uF. This is especially true with things like +12V2A and +5V 2A PSUs. However, the simple CL circuit isn't so happy with that, and it may cause a small amount of "hunting" in the output voltage, as well as possible overheating of the device with the 33 ohm emitter resistor as the large electrolytics "form." A cap across the input could be worthwhile if the JLH is some distance from the voltage.source . I use 7.5A mains cable to help overcome this problem, and I have thought about using a 100nF at the input for that reason., You appear to be mixing up your CCS and CL again ? When the CL is used at lower currents, it does appear to give a small additional filtering effect. I get better results with higher current supplies by using higher input filter capacitor values in the PSU itself than commonly seen ( e.g. 2 x 4,700uF in parallel) . Reducing the amount of work needed to remove minor 100HZ ripple appears to let the JLH work better with HF rubbish. I have seen these improvements on a C.R.O. with my +12V 2A and +5V 2A PSU. I think that Allan and Jeff may have done similar with their high current supplies too ? I would welcome any thoughts by Marc, Shaun,Allan and others on this subject. Kind Regards Alex
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 2:10:13 GMT
Hi Greg Just a couple of quick thoughts. Most LEDS these days have a forward voltage of > 1.8V. We would need to take that into account. LEDs also have a fairly low reverse voltage rating. Of course. Red LED Vf = 1.7V maths is easy. Any other issue? My experience is that above 12V there is no audible advantage to replacing the 10K with a far more expensive CRD. Above 24V or so it may be worthwhile increasing the value of the 10K resistor in proportion to the increase of the supply voltage. What 10k resistor? If R11 if agree. When using the JLH at higher currents than the original design, there is a definite SQ advantage in going to parallel 4,700uF capacitors instead of the original 2,200uF. This is especially true with things like +12V2A and +5V 2A PSUs. However, the simple CL circuit isn't so happy with that, and it may cause a small amount of "hunting" in the output voltage, as well as possible overheating of the device with the 33 ohm emitter resistor as the large electrolytics "form." I'll just what to concentrate on low power situations at the moment using the standard JLH Ripple Eater. A cap across the input could be worthwhile if the JLH is some distance from the voltage.source . I use 7.5A mains cable to help overcome this problem, and I have thought about using a 100nF at the input for that reason. I am just trying to see if it worth allowing for a cap on future PCBs. You appear to be mixing up your CCS and CL again ? How so? It seems very clear in my head. The topic at the moment is the CL. When I was referring to a CCS it was in reference of replacing R11. When the CL is used at lower currents, it does appear to give a small additional filtering effect. My thoughts (not experience) also. We need to determine when to get rid of it. 100mA 200mA 500ma? I would welcome any thoughts by Marc, Shaun,Allan and others on this subject. . Yes, I need more mentors. regards Greg
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 3:41:01 GMT
Hi Greg A red LED has a less well defined forward voltage, as well as higher leakage and a fairly low reverse voltage rating. I am not saying that this will matter here, but should be taken into consideration when replacing 2 Si diodes in series.. "The original CL has a output resistance in the order of 500K. This output resistance reduces when the CL is tuned for 5 volts. ? ? ? The CL is normally a simple transistor switch in saturated mode.With less current through it, and lower rail voltages the bias current doesn't need to be as high. " I'll just what to concentrate on low power situations at the moment using the standard JLH Ripple Eater" These days, most people are using the JLH with the CL bypassed in order to increase the effective simulated capacitance with parallel 4,700uFs and at possibly higher current than the original design. Some applications of the JLH, actually sound better with the CL in circuit, such as the Class A preamp/HA., so you can't generalise as to what current should see the CL section bypassed If it ain't broke, don't fix it ? By all means try further refinements, but if we do we may need to do some objective testing with a DC supply that has a superimposed 100KHZ triangular waveform as David did, and look closely at the output with a CRO, possibly using a 10 x low noise preamp. I had to use a 10 x low noise preamp with my CRO in order to see any meaningful residual from the output of the Class A preamp which uses 2 JLHs. Regards Alex
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 6:15:54 GMT
Hi Greg Further thoughts. The original CL design with 6R8 emitter resistor starts to fold back current around 88mA.(depends on VF of both series diodes and VBE of the CL transistor.) With 1.8V Red LED (typical) and 6R8 emitter resistor ,the CL starts to fold back current at about 176mA and there will be an unacceptably high voltage drop through the JLH well before that happens. Regards Alex
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 7:40:20 GMT
Hi Alex, Thanks for putting in a big effort to answer my questions. I will digest the information and I will be asking more questions!! Just to be clear, when I talk about the CL I am talking about the complete circuit D2, D3, Q6, R11 and R12. Adjusting R11 and R12 changes the maximum current of the CL. R11 has a small effect on D2 and D3 Vf and the Vbe of C6. users.tpg.com.au/gerskine/greg/media/jlh%20regulator%20dual%20mk%207%20sch%201.pdfregards Greg
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 8:21:08 GMT
This is the original circuit that Greg is discussing here. Decreasing the value of R11 doesn't alter the current capabilities. It ensures that if a higher current limit is required, that the CL transistor will remain close to saturation in order to avoid aextra voltage drop across it's collector and emitter junction. At lower currents , provided the CL device remains in saturated mode, the total voltage drop across the JLH may be as low as 300mV. Current Version. Click on small image, then click again on larger image.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 21:52:48 GMT
I see Q1&2 have been replaced with the 2SC2240, something I'd not noticed before.
What's the advantage of the change here and is it suitable for all applications over the BC549C?
EDIT; just found the noise figure is typical 2dB instead 4dB...
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 7, 2013 22:04:00 GMT
I see Q1&2 have been replaced with the 2SC2240, something I'd not noticed before. What's the advantage of the change here and is it suitable for all applications over the BC549C? EDIT; just found the noise figure is typical 2dB instead 4dB... Hi Chris OOPS ! I posted the wrong schematic from what was already saved in ImageShack. You will see at the bottom that this is a High Voltage version. (+-55V) as used by Allan from Sydney for the front end of one of his amplifiers. For lower voltages BC549C/BC559C are fine, but BC550C and BC560C will cover most situations and are much cheaper in larger quantities. Kind Regards Alex.. P>S. Anything that you would like to see with a new generation of JLH PCBs, or any suggestions for a further improvement please post in this thread. Please bear in mind though, that large value SMD electrolytic capacitors of more than a few volts rating, and with adequate ripple ratings may not be readily available, if at all. IIRC, Greg has now made and distributed a total of around 800 JLH PCBs in different versions, for use world wide.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 8, 2013 6:13:52 GMT
Decreasing the value of R11 doesn't alter the current capabilities. It ensures that if a higher current limit is required, that the CL transistor will remain close to saturation in order to avoid aextra voltage drop across it's collector and emitter junction. At lower currents , provided the CL device remains in saturated mode, the total voltage drop across the JLH may be as low as 300mV. Hi Alex, Again thanks for your input. I think I am learning something, (maybe not). I think, changing R11 changes the current (If) flowing diodes D2, D3. Changing the current though D2 and D3 changes their Vf. Changing their Vf changes the required value of R12. Setting diode current (If) to 5mA gives a Vf of 0.6 to 0.7V which is what we usually use for hand calculations. Your 1k8 value for R11 nicely sets the diode current to roughly 5 or 6 mA. It probably changes Vbe of Q6 slightly though I haven't checked this? At the end of the day, obviously R11 doesn't effect the maximum current as much as R12. regards Greg
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 8, 2013 6:34:23 GMT
Decreasing the value of R11 doesn't alter the current capabilities. It ensures that if a higher current limit is required, that the CL transistor will remain close to saturation in order to avoid aextra voltage drop across it's collector and emitter junction. At lower currents , provided the CL device remains in saturated mode, the total voltage drop across the JLH may be as low as 300mV. Hi Alex, Again thanks for your input. I think I am learning something, (maybe not). I think, changing R11 changes the current (If) flowing diodes D2, D3. Changing the current though D2 and D3 changes their Vf. Changing their Vf changes the required value of R12. Setting diode current (If) to 5mA gives a Vf of 0.6 to 0.7V which is what we usually use for hand calculations. Your 1k8 value for R11 nicely sets the diode current to roughly 5 or 6 mA. It probably changes Vbe of Q6 slightly though I haven't checked this? At the end of the day, obviously R11 doesn't effect the maximum current as much as R12. regards Greg Hi Greg There is no current through these diodes until current limiting starts. For them to conduct you add VBE of the CL device and the voltage drop across the emitter resistor. This then needs to be in the vicinity of 1.2V before the diodes can conduct. With the original circuit , that means the load will need to be around 88mA or more. In the lower current JLH as published with 4.7K and 6.8 ohms,, it could be interesting to try a J510 or J511 CRD ( 3or 4,7mA) in place of R11. This may result in a slightly more stable supply. With higher currents, the base current will decrease due to the higher voltage drop across it's emitter resistor. This MAY keep the base current constant ? Anybody , feel free to correct me here, if in error. Regards Alex
|
|