I'm sitting here in my lab with a 289, a signal generator and an oscilloscope.
It's not totally clear from your description, but I'm guessing that what you're trying to do is measure the switching frequency of the load connected to your power supply by measuring the frequency of the switching noise that appears superimposed on the 4V dc supply.
With the 289 set in DCV mode and the frequency display enabled, it seems that the 289 requires a very large signal in order to register a frequency measurement. A 4 kHz square wave oscillating between -2V and +2V does the trick, though there's also a dependency on the dc voltage offset, and the duty cycle measurement is still way off. I suspect there's a low pass filter in the meter which is used to give a steady DC reading in the presence of noise, and this has the effect of dramatically reducing the sensitivity of the frequency measurement too. I actually find the idea of measuring frequency when in DC mode a bit odd.
However, what you're trying to measure is the AC component of the signal coming from the power supply, so you want the meter on ACV. In this mode there's a high pass filter instead of a low pass filter, so the AC signal you're looking to measure isn't attenuated.
In ACV mode the meter's response is better. I can measure a 4kHz square wave all the way down to 0.5Vpp @ 4kHz, though this is probably still a much larger signal than the amount of noise you'll have on your power supply.
It's still not much use for measuring pulses with a short duty cycle, though. At 4kHz and 1Vpp it doesn't register anything nearly correct until the duty cycle is 23% or more, and it's similarly useless if the duty cycle is greater than 80%. Larger signals result in better performance, but your signal will be small.
In short, I'm afraid the 289 isn't a good tool for the job when used in this way, and you'd really need an oscilloscope. But there's a better way to use the meter to make the measurement you're after.
Put it in current mode and wire the 289 in series with your load. In current mode the 289 records the correct switching frequency at only a couple of mA, and duty cycle measurements work fine too. Moreover, you're actually measuring the duty cycle of your load rather than just the switching spikes which it imposes on the power supply.