calculating high-pass filter for these components

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I understand that the formula for determining f3 for a high-pass, CR filter is 1/(2 Pi RC)
(please correct if that is not right)

The sample given in Horowitz and Hill does not have the resistor in series with the capacitor, i.e. the 1k R701L. That is what is confusing me.

Q: What would be used for R to solve the equation?

Q: R can not be R701L alone, because the result would be too high. But does that resistor have any effect on the f3 value?

Q: Is there a low-pass filter here too? What is the function of C702?


thanks in advance...
 

Attachments

  • CR filter sample calculation.jpg
    CR filter sample calculation.jpg
    44.2 KB · Views: 141
Moderator
Joined 2011
I understand that the formula for determining f3 for a high-pass, CR filter is 1/(2 Pi RC)
The sample given in Horowitz and Hill does not have the resistor in series with the capacitor, i.e. the 1k R701L. That is what is confusing me.
Q: What would be used for R to solve the equation?
Q: Is there a low-pass filter here too? What is the function of C702?

The low pass filter is the 1k and 10pF.
The high pass filter is the 0.47uF and 220k.
It's a little confusing when they're combined as shown.
The low filter is far apart in frequency from the high filter,
so you can neglect one filter's parts when calculating the other filter.
 
Last edited:
Thank you rayma

I calculate the high-pass f3 to be around 1.5Hz

I have read elsewhere that high-pass to amp inputs are generally kept below 5Hz, so this example is well below that already.


Would there be any harm in increasing C701L to 1uF? That would decrease f3 to around .72Hz
 
I recommend that the passband of the amplifier be determined by the two passive input filters.

To ensure this actually happens one needs to check what other roll offs are occurring inside the amplifier.
The NFB upper and lower leg resistors usually have capacitances attached that create a filtering/roll off effect.
Calculate what these roll off frequencies are and adjust your input filters so that they have a narrower passband than the NFB filters.

If you have a scope you can apply a test signal to the input after the input filters, (but be careful, some amps oscillate and blow up if you attach anything to the input -IN pin).
Or cripple the input filters by changing the capacitances and apply the test signal to the conventional input and then see at what frequencies the amp starts to roll off.

A reduction in the 2.83Vac (8Vpp) output to 90% (2.5Vac, 7.2Vpp) is ~-1dB with ref to the flat passband voltage. If you double/halve your F-1dB frequency, you get a very rough approximation to the F-3dB frequency.
 
Last edited:
Yes, the formula works for both the high pass filter CR and for the low pass filter RC.
These are passive single pole filters.

Your 0u47F + 220k has F-3dB= 1.54Hz. The F-1dB is ~3Hz. A scope would show a 10% drop in output at ~3Hz.
The RC time constant for your 0u47F + 220k is 0.1034seconds (=103ms). This is a very convenient way to compare filters.
eg. the FM pre-emphasis is either 50us, or 75us. RIAA & NAB are also done in time constants.
 
Last edited:
rayma said:
The low pass filter is the 1k and 10pF.
The high pass filter is the 0.47uF and 220k.
It's not quite as simple as that. There may be some Miller capacitance to add to the 10pF. There may be some BJT input resistance to add in parallel with the 220k. In each case it depends on the rest of the circuit, so the effect could be negligible or significant. Also, there may be some source resistance to add to the 1k.
 
Thanks for the additional thoughts guys.

Q: 1uF OK for the input cap here or no? Not enough of the scat showing to determine?


Q: re time constants, is there a benefit in considering filters by their time constants? Is there some overall strategy in the design of an amplifier whereby the time constants of all the various filters are coordinated in some way?
 
Moderator
Joined 2011
It's not quite as simple as that. There may be some Miller capacitance to add to the 10pF. There may be some BJT input resistance to add in parallel with the 220k. In each case it depends on the rest of the circuit, so the effect could be negligible or significant. Also, there may be some source resistance to add to the 1k.

Just trying to keep things simple for a beginner.
 
re time constants, is there a benefit in considering filters by their time constants?
Is there some overall strategy in the design of an amplifier whereby the time constants of all the various filters are coordinated in some way?

Yes, since the time constants are inversely proportional to the corner frequencies (called poles and zeros),
they're just another way of looking at the same information. A time constant (for example) is (RxC) with units of seconds,
while the corner frequency is 1/(2xPixRxC) with units of Hertz (cycles per second).

Usually there should be one dominant time constant (the largest one) in the circuit for good high frequency response and stability.
Or, you could say there should be one dominant pole (high frequency roll off point) that is much lower in frequency
than others in the circuit, by say ten times. This keeps the phase shift low. Often this is set by the input filter low pass filter.

Similarly, at low frequencies there should be one dominant pole (low frequency rolloff) that is much larger (higher in frequency)
than others in the system, by say ten times, for good stability and low frequency response. Or, you could say there should be one dominant
time constant much smaller than others in the circuit. This keeps the phase shift low. Often this is set by the input high pass filter.

There are also various techniques to modify the phase shifts in a circuit to improve stability and extend the response.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.