I recently came across a user-modified Philips (Magnavox) CD player (from the late 1980s) in which several electrolytic caps had been replaced.
The original (Philips schematic) values are 33uF/16v. The user had replaced them with Panasonic FM 100uF/16v.
These are: electro found after main LM7805 regulator (at output) and electros found on Vdd (power) rails of various ICs.
Questions:
---Why did Philips select that particular value (33uF)?
---Since the modded values are 3x higher (100uF), could this limit the performance of the IC devices?
Note: I have seen Philips use 47uF for these same ICs in other CD players of similar vintage.
Thx!
The original (Philips schematic) values are 33uF/16v. The user had replaced them with Panasonic FM 100uF/16v.
These are: electro found after main LM7805 regulator (at output) and electros found on Vdd (power) rails of various ICs.
Questions:
---Why did Philips select that particular value (33uF)?
---Since the modded values are 3x higher (100uF), could this limit the performance of the IC devices?
Note: I have seen Philips use 47uF for these same ICs in other CD players of similar vintage.
Thx!
FM's are very good caps and low ESR. I'd be tempted to change the one on the output of the 7805 to something with higher ESR as the low ESR could upset the reg. The others I would leave in, they're probably better than the originals in those positions.
What about their value?
I think I've heard increasing the values too high can lead to muddy bass, etc. Esp. at the regulator.
In any case, I'm sure Philips had important reasons for selecting the values they did.
Yeah, I know FM are good stuff ... but my query was about the difference in values between stock and mod. Thoughts?
I think I've heard increasing the values too high can lead to muddy bass, etc. Esp. at the regulator.
In any case, I'm sure Philips had important reasons for selecting the values they did.
Using anything larger than 100uF in power supply rails' decoupling in digital audio players is not required.
What works nicely is around 100uF decoupling before a voltage regulator (and then a cap of a value suggested by the manufacturer at the Vout pin), followed by between 2.2uF and 4.7uF decoupling few centimeters away from Vcc / Vdd pins, followed by 0.01 to 0.1uF decoupling at the power supply pins. This will provide a (reasonably) low impedance, very fast power supply rails to correctly supply the current at very large frequency span even to analog stages used in digital devices.
Another approach is to use local regulators at each pin, in which case a small 2.2uF with 0.01 to 0.1uF for digital devices is exactly what's required.
Shunt regulators will provide an optimum output impedance to analog stages, in which case the 10uF / 0.1uF combo is all that is required even for very low frequencies (below 20Hz).
What works nicely is around 100uF decoupling before a voltage regulator (and then a cap of a value suggested by the manufacturer at the Vout pin), followed by between 2.2uF and 4.7uF decoupling few centimeters away from Vcc / Vdd pins, followed by 0.01 to 0.1uF decoupling at the power supply pins. This will provide a (reasonably) low impedance, very fast power supply rails to correctly supply the current at very large frequency span even to analog stages used in digital devices.
Another approach is to use local regulators at each pin, in which case a small 2.2uF with 0.01 to 0.1uF for digital devices is exactly what's required.
Shunt regulators will provide an optimum output impedance to analog stages, in which case the 10uF / 0.1uF combo is all that is required even for very low frequencies (below 20Hz).
Whoa ... you mean right after the primary diode bridge? isn't that supposed to be in the several-thousand uF range?
This advice sounds reasonable.
However, Philips does not follow this strategy (TTBOMK).
Are there any mainstream manufs. (Sony, Denon, etc.) that do follow this strategy?
How about non-mainstream. That is, audiophile and high end. I must admit i haven't browsed thru schematics of very many non-mainstreams, like Naim, Arcam, et. al.
While on the general topic of vintage Philips CD players, I will note another curiosity: the use of a lo-value resistor (2-10 ohm) at the power rails (Vdd pins) of ICs (opamps, digital ICs, etc.).
While it's certain that these R's reduce noise, I'm not sure how they affect power delivery? I assume we want fast and clean power? If the power was already pretty clean (e.g. from a souped-up PSU section), would removing these R's help?
I have seen some DIY projects replace the R's with ferrrite beads.
While it's certain that these R's reduce noise, I'm not sure how they affect power delivery? I assume we want fast and clean power? If the power was already pretty clean (e.g. from a souped-up PSU section), would removing these R's help?
I have seen some DIY projects replace the R's with ferrrite beads.
Last edited:
Both approaches work at removing very high ingress/egress frequency rubbish... but cause unpredictable results with regard to overall power supply rail impedance/inductance. This makes it very hard to correctly calculate the required capacitance along the whole power supply rail.
Philips used to use low value fusible resistors like they were going out of fashion, not just in audio but TV/Video products as well.
I would leave them, they are a designed in safety feature and yes, they will help reduce noise and random fluctuations on the rails.
The 33uF caps were obviously deemed most suitable in the design phase, as such I would stick with the original or perhaps 47uF. Some players used 33uF in locations on the servo board (such as around the laser drive transistor) and I would stick to the design value here.
I would leave them, they are a designed in safety feature and yes, they will help reduce noise and random fluctuations on the rails.
The 33uF caps were obviously deemed most suitable in the design phase, as such I would stick with the original or perhaps 47uF. Some players used 33uF in locations on the servo board (such as around the laser drive transistor) and I would stick to the design value here.
Actually doing the some modding with a Hafler SE-150, based on a Philips mainboard (similar to Philips CD-582). After checking the schematics, i will only change the capacitors before the regulators (some with higher values) and 47µF after the regulators and replace the opamps (LM833) with OPA2604 and the bipolar Nichicon ES at output with new ones.
Also find in the net an idea to supply each main digital circuit with separate regulators for maximum decoupling.
BR
Günni
Also find in the net an idea to supply each main digital circuit with separate regulators for maximum decoupling.
BR
Günni
Be careful where you get the OPA2604 from as these have been out of production for some time due to problems with production.
thanks Mooly for that advice.
The opamp locations will have sockets, so i have the possibilities to play a little with different types, like NE5532 AN, LME49720, OPA2134 or LM4562 or back again to the original equipped.
BR
Günni
The opamp locations will have sockets, so i have the possibilities to play a little with different types, like NE5532 AN, LME49720, OPA2134 or LM4562 or back again to the original equipped.
BR
Günni
Take a look ...
Note the 6800uF cap before the 7805 reg. Yes, I realize that:
-- this cap is also meant for that unregulated 10v that goes to the Focus Drive opamp
-- Your mods suggest further decoupling at the IC. E.g. 2.2uF - 4.7uF.
But still, 6800uF is several orders of magnitude higher than "100uF decoupling before a voltage regulator".
Alas, I have not done my own experiments to confirm/deny any benefits to your suggested mods. I'd assumed Philips knew what they were doing. Also, that rectifier topology (diode bridge + large cap + regulator) is ubiquitous in electronics.
Take a look at this segment from a Philips CD-482 schematic ...
An externally hosted image should be here but it was not working when we last tested it.
Note the 6800uF cap before the 7805 reg. Yes, I realize that:
-- this cap is also meant for that unregulated 10v that goes to the Focus Drive opamp
-- Your mods suggest further decoupling at the IC. E.g. 2.2uF - 4.7uF.
But still, 6800uF is several orders of magnitude higher than "100uF decoupling before a voltage regulator".
Alas, I have not done my own experiments to confirm/deny any benefits to your suggested mods. I'd assumed Philips knew what they were doing. Also, that rectifier topology (diode bridge + large cap + regulator) is ubiquitous in electronics.
Unless ...
I've been going over some Arcam schematics and they, for certain regs, contain both these cap values (one of each) between the diode bridge and regulator.
I emphasize: for certain regs. For others, it's the same situation as the Philips schematic in my last post.
Refs:
Arcam FMJ-CD23 - HiFi Engine
Unless you meant that several-thousand uF cap and (in parallel with) "100uF decoupling before a voltage regulator".
I've been going over some Arcam schematics and they, for certain regs, contain both these cap values (one of each) between the diode bridge and regulator.
I emphasize: for certain regs. For others, it's the same situation as the Philips schematic in my last post.
Refs:
Arcam FMJ-CD23 - HiFi Engine
Last edited:
It depends on (at least) the amount of current being pulled, and how high above the regulator voltage the secondary is. If you don't flatten the ripple enough before the regulator, the dips may be below its headroom voltage.
On the other hand, if you increase the capacitance too much, you'll put a larger load on the transformer, which may be more than it's rated for.
If you don't have all the data to model it in SPICE, I suspect that anything between 100% and 150% of the original value would do fine. (Make sure that the voltage and ripple current are also at least as much as the originals. I don't think they have the same issues with going too high.)
On the other hand, if you increase the capacitance too much, you'll put a larger load on the transformer, which may be more than it's rated for.
If you don't have all the data to model it in SPICE, I suspect that anything between 100% and 150% of the original value would do fine. (Make sure that the voltage and ripple current are also at least as much as the originals. I don't think they have the same issues with going too high.)
It does indeed look like it's awfully close to that reg.... on that screen-capture.
Using hifiengine's Manual Library, I've been poring over myriad SMs (schematics) from Rotel, Arcam, Cambridge, etc., and have found cap/regulator strategies to vary. So much, in fact, that the final section values are either arbitrary or highly tweaked.
Even presence of that small-value ceramic (or plastic) bypass cap (parallel with electro) is inconsistent.
The only "constant" design is: diode bridge ---> large-value electro cap ---> regulator input.
Even presence of that small-value ceramic (or plastic) bypass cap (parallel with electro) is inconsistent.
The only "constant" design is: diode bridge ---> large-value electro cap ---> regulator input.
One issue with electrolytic capacitors is that they are inductive at higher frequencies, so for digital supplies, I try and use solid polymer types. A potential problem with introducing these however, is that they have an insanely low esr, and can ring. I use no larger than 47f to 100uf, to help with that, and to me, work way better than electrolytic.
Solids are de reguer in "better" computer motherboards and PSUs. I think they are superior: not in performance; rather, reliability and durability
Shifting gears back to regulators and their electro caps... In the schema I posted above ...
An externally hosted image should be here but it was not working when we last tested it.
...did anyone notice the asymmetric input capacitance:
220uF ---> 7815 ---> 47u ---> +15v
470uF ---> 7915 ---> 47u ---> -15v
Among other rails, these rails feed the same I/V and LPF opamps.
Correct me if I'm wrong ... but the asymmetry is due to the inherent difference in design between 78xx and 79xx regulators??
- Status
- Not open for further replies.