Slow ramping of the CELSIUS storage ring has been utilized to measure the yield of charged pions in proton and heavy ion induced collisions with continuously varying beam energy. Boltzmann-Uehling-Uhlenbeck predictions, including Fermi momenta of nucleons in nuclei, follow the general shape of the p-nucleus excitation functions quite well except for a general overestimation of the backward emission. For heavy ion reactions the calculated yield also falls off faster with decreasing beam energy than the data. No statistically significant narrow resonances are observed. [S0031-9007(97)03152-9] PACS numbers: 25.40.Ve, 24.10.Nz, Pions can be produced in hadron-nucleus and nucleusnucleus collisions, also at energies well below the free nucleon-nucleon ͑NN͒ threshold through the collective interaction between several nucleons or through the boost from the internal nucleon momentum. Suggested production mechanisms range from first chance NN scattering [1,2], full cascade prescriptions [3,4], dynamical mean-field 1 NN collision equations [5-8], predominantly for reactions around and above the NN threshold, to cluster-cluster interactions [9] and fully collective pionic fusion models [10-12], for reactions close to the absolute (collective) threshold. The strong rescattering when pions or deltas propagate through the nuclei must, of course, be introduced in all models.Many of these models predict the overall features of pion emission well but a detailed selection among them is hampered by the lack of systematic data. By combining slow ramping operation of the CELSIUS storage ring with the range telescope technique, complete excitation functions for p 6 emission over wide ranges of beam energy can be measured-from the absolute (collective) threshold to well above the (free) NN threshold.Protons and fully stripped Ne ions were injected into the CELSIUS storage ring [13], accelerated up to the start energy, stored during slow ramping of the magnets with gas-jet target in operation until the final energy was reached, and then finally dumped. The cycle time, 2-5 min, was governed by the requirement that at least 1 3 of the stored ions should remain. Data were collected continuously during 70-250 s. The start time for slow ramping, the time when the event trigger appeared, and the beam frequency at that time were stored to provide the collision energy. No electron cooling of the beams was introduced. Data from two ramp cycles, one at low energy (169-270 MeV for p and 50A 120A MeV for Ne) and one high energy (250-500 MeV and 100A 400A MeV), were put together. The reproduction of the cycles was very good and data could be added for several hours without increasing the collision energy dispersion (, 0.8 MeV for proton beams and , 0.9A MeV for Ne beams). Gas-jet target thicknesses of between 10 13 and 10 14 atoms͞cm 2 for N, Ar, Kr, and Xe gases were employed.Five 9-element plastic range telescopes were mounted outside thin steel windows in angular positions from 20 ± to 120 ± . These telescopes [14] have increasing scintillator thi...